go back to using @dataclass since they can be inspected

rebased jpt/readme branch

.github/workflows/publish.yaml

@@ -46,7 +46,7 @@ jobs:
     needs: [ build ]
     environment:
       name: pypi
-      url: https://pypi.org/p/dailyai
+      url: https://pypi.org/p/pipecat-ai
     permissions:
       id-token: write
     steps:
@@ -60,3 +60,25 @@ jobs:
         with:
           verbose: true
           print-hash: true
+
+  publish-to-test-pypi:
+    name: "Publish to Test PyPI"
+    runs-on: ubuntu-latest
+    needs: [ build ]
+    environment:
+      name: testpypi
+      url: https://pypi.org/p/pipecat-ai
+    permissions:
+      id-token: write
+    steps:
+      - name: Download wheels
+        uses: actions/download-artifact@v4
+        with:
+          name: wheels
+          path: ./dist
+      - name: Publish to PyPI
+        uses: pypa/gh-action-pypi-publish@release/v1
+        with:
+          verbose: true
+          print-hash: true
+          repository-url: https://test.pypi.org/legacy/

.github/workflows/publish_test.yaml

@@ -15,6 +15,8 @@ jobs:
         uses: actions/checkout@v4
         with:
           ref: ${{ github.event.inputs.gitref }}
+          fetch-tags: true
+          fetch-depth: 100
       - name: Set up Python
         id: setup_python
         uses: actions/setup-python@v4
@@ -35,21 +37,23 @@ jobs:
       - name: Upload wheels
         uses: actions/upload-artifact@v4
         with:
+          name: wheels
           path: ./dist
 
   publish-to-pypi:
-    name: "Test publish to PyPI"
+    name: "Publish to Test PyPI"
     runs-on: ubuntu-latest
     needs: [ build ]
     environment:
-      name: pypi
-      url: https://pypi.org/p/dailyai
+      name: testpypi
+      url: https://pypi.org/p/pipecat-ai
     permissions:
       id-token: write
     steps:
       - name: Download wheels
         uses: actions/download-artifact@v4
         with:
+          name: wheels
           path: ./dist
       - name: Publish to PyPI
         uses: pypa/gh-action-pypi-publish@release/v1

.gitignore

@@ -3,6 +3,7 @@ env/
 __pycache__/
 *~
 venv
+.venv
 #*#
 
 # Distribution / packaging

README.md

@@ -1,77 +1,130 @@
-# dailyai — an open source framework for real-time, multi-modal, conversational AI applications
+<div align="center">
+ <img alt="pipecat" width="300px" height="auto" src="pipecat.png">
+</div>
+
+# Pipecat
+
+[![PyPI](https://img.shields.io/pypi/v/pipecat-ai)](https://pypi.org/project/pipecat-ai) [![Discord](https://img.shields.io/discord/1239284677165056021
+)](https://discord.gg/pipecat)
+
+`pipecat` is a framework for building voice (and multimodal) conversational agents. Things like personal coaches, meeting assistants, story-telling toys for kids, customer support bots, and snarky social companions.
 
 Build things like this:
 
 [![AI-powered voice patient intake for healthcare](https://img.youtube.com/vi/lDevgsp9vn0/0.jpg)](https://www.youtube.com/watch?v=lDevgsp9vn0)
 
-**`dailyai` started as a toolkit for implementing generative AI voice bots.** Things like personal coaches, meeting assistants, story-telling toys for kids, customer support bots, and snarky social companions.
+## Getting started with voice agents
 
-In 2023 a *lot* of us got excited about the possibility of having open-ended conversations with LLMs. It became clear pretty quickly that we were all solving the same [low-level problems](https://www.daily.co/blog/how-to-talk-to-an-llm-with-your-voice/):
-- low-latency, reliable audio transport
-- echo cancellation
-- phrase endpointing (knowing when the bot should respond to human speech)
-- interruptibility
-- writing clean code to stream data through "pipelines" of speech-to-text, LLM inference, and text-to-speech models
+You can get started with Pipecat running on your local machine, then move your agent processes to the cloud when you’re ready. You can also add a telephone number, image output, video input, use different LLMs, and more.
 
-As our applications expanded to include additional things like image generation, function calling, and vision models, we started to think about what a complete framework for these kinds of apps could look like.
-
-Today, `dailyai` is:
-
-1. a set of code building blocks for interacting with generative AI services and creating low-latency, interruptible data pipelines that use multiple services
-2. transport services that moves audio, video, and events across the Internet
-3. implementations of specific generative AI services
-
-Currently implemented services:
-- Speech-to-text
-  - Deepgram
-  - Whisper
-- LLMs
-  - Azure
-  - OpenAI
-- Image generation
-  - Azure
-  - Fal
-  - OpenAI
-- Text-to-speech
-  - Azure
-  - Deepgram
-  - ElevenLabs
-- Transport
-  - Daily
-  - Local (in progress, intended as a quick start example service)
-
-If you'd like to [implement a service]((https://github.com/daily-co/daily-ai-sdk/tree/main/src/dailyai/services)), we welcome PRs! Our goal is to support lots of services in all of the above categories, plus new categories (like real-time video) as they emerge.
-
-## Getting started
-
-Today, the easiest way to get started with `dailyai` is to use [Daily](https://www.daily.co/) as your transport service. This toolkit started life as an internal SDK at Daily and millions of minutes of AI conversation have been served using it and its earlier prototype incarnations. (The [transport base class](https://github.com/daily-co/daily-ai-sdk/blob/main/src/dailyai/transports/abstract_transport.py) is easy to extend, though, so feel free to submit PRs if you'd like to implement another transport service.)
-
-```
+```shell
 # install the module
-pip install dailyai
+pip install pipecat-ai
 
 # set up an .env file with API keys
 cp dot-env.template .env
 ```
 
-By default, in order to minimize dependencies, only the basic framework functionality is available. Some third-party AI services require additional
-dependencies that you can install with:
+By default, in order to minimize dependencies, only the basic framework functionality is available. Some third-party AI services require additional dependencies that you can install with:
 
-```
-pip install dailyai[option,...]
+```shell
+pip install "pipecat-ai[option,...]"
 ```
 
 Your project may or may not need these, so they're made available as optional requirements. Here is a list:
 
-- **AI services**: `anthropic`, `azure`, `fal`, `openai`, `playht`, `silero`, `whisper`
+- **AI services**: `anthropic`, `azure`, `fal`, `moondream`, `openai`, `playht`, `silero`, `whisper`
 - **Transports**: `daily`, `local`, `websocket`
 
+## A simple voice agent running locally
+
+If you’re doing AI-related stuff, you probably have an OpenAI API key.
+
+To generate voice output, one service that’s easy to get started with is ElevenLabs. If you don’t already have an ElevenLabs developer account, you can sign up for one [here].
+
+So let’s run a really simple agent that’s just a GPT-4 prompt, wired up to voice input and speaker output.
+
+You can change the prompt, in the code. The current prompt is “Tell me something interesting about the Roman Empire.”
+
+`cd examples/getting-started` to run the following examples …
+
+```shell
+# Talk to a local pipecat process with your voice. Specify GPT-4 as the LLM.
+
+export OPENAI_API_KEY=...
+export ELEVENLABS_API_KEY=...
+python ./local-mic.py | ./pipecat-pipes-gpt-4.py | ./local-speaker.py
+```
+
+## WebSockets instead of pipes
+
+To run your agent in the cloud, you can switch the Pipecat transport layer to use a WebSocket instead of Unix pipes.
+
+```shell
+# Talk to a local pipecat process with your voice. Specify GPT-4 as the LLM.
+
+export OPENAI_API_KEY=...
+export ELEVENLABS_API_KEY=...
+python ./local-mic-and-speaker-wss.py wss://localhost:8088
+```
+
+## WebRTC for production use
+
+WebSockets are fine for server-to-server communication or for initial development. But for production use, you’ll need client-server audio to use a protocol designed for real-time media transport. (For an explanation of the difference between WebSockets and WebRTC, see [this post.])
+
+One way to get up and running quickly with WebRTC is to sign up for a Daily developer account. Daily gives you SDKs and global infrastructure for audio (and video) routing. Every account gets 10,000 audio/video/transcription minutes free each month.
+
+Sign up [here](https://dashboard.daily.co/u/signup) and [create a room](https://docs.daily.co/reference/rest-api/rooms) in the developer Dashboard. Then run the examples, this time connecting via WebRTC instead of a WebSocket.
+
+```shell
+# 1. Run the pipecat process. Provide your Daily API key and a Daily room
+export DAILY_API_KEY=...
+export OPENAI_API_KEY=...
+export ELEVENLABS_API_KEY=...
+python pipecat-daily-gpt-4.py --daily-room https://example.daily.co/pipecat
+
+# 2. Visit the Daily room link in any web browser to talk to the pipecat process.
+#    You'll want to use a Daily SDK to embed the client-side code into your own
+#    app. But visiting the room URL in a browser is a quick way to start building
+#    agents because you can focus on just the agent code at first.
+open -a "Google Chrome" https://example.daily.co/pipecat
+```
+
+## Deploy your agent to the cloud
+Now that you’ve decoupled client and server, and have a Pipecat process that can run anywhere you can run Python, you can deploy this example agent to the cloud.
+
+`TBC`
+
+## Taking it further
+
+### Add a telephone number
+Daily supports telephone connections in addition to WebRTC streams. You can add a telephone number to your Daily room with the following REST API call. Once you’ve done that, you can call your agent on the phone.
+
+You’ll need to add a credit card to your Daily account to enable telephone numbers.
+
+`TBC`
+
+
+### Add image output
+
+Daily supports telephone connections in addition to WebRTC streams. You can add a telephone number to your Daily room with the following REST API call. Once you’ve done that, you can call your agent on the phone.
+
+You’ll need to add a credit card to your Daily account to enable telephone numbers.
+
+`TBC`
+
+### Add video output
+
+
+`TBC`
+
+
 ## Code examples
 
 There are two directories of examples:
 
-- [foundational](https://github.com/daily-co/daily-ai-sdk/tree/main/examples/foundational) — demos that build on each other, introducing one or two concepts at a time
-- [starter apps](https://github.com/daily-co/daily-ai-sdk/tree/main/examples/starter-apps) — complete applications that you can use as starting points for development
+- [foundational](https://github.com/daily-co/pipecat/tree/main/examples/foundational) — examples that build on each other, introducing one or two concepts at a time
+- [starter apps](https://github.com/daily-co/pipecat/tree/main/examples/starter-apps) — complete applications that you can use as starting points for development
 
 Before running the examples you need to install the dependencies (which will install all the dependencies to run all of the examples):
 
@@ -97,7 +150,7 @@ source venv/bin/activate
 From the root of this repo, run the following:
 
 ```
-pip install -r {env}-requirements.txt -r dev-requirements.txt
+pip install -r dev-requirements.txt -r {env}-requirements.txt
 python -m build
 ```
 

dev-requirements.txt

@@ -1,6 +1,6 @@
-autopep8==2.0.4
-build==1.0.3
-pip-tools==7.4.1
-pytest==8.1.1
-setuptools==69.2.0
-setuptools_scm==8.0.4
+autopep8~=2.1.0
+build~=1.2.1
+pip-tools~=7.4.1
+pytest~=8.2.0
+setuptools~=69.5.1
+setuptools_scm~=8.1.0

docs/README.md

@@ -1,8 +1,8 @@
-# Daily AI SDK Docs
+# Pipecat Docs
 
 ## [Architecture Overview](architecture.md)
 
-Learn about the thinking behind the SDK's design.
+Learn about the thinking behind the framework's design.
 
 ## [A Frame's Progress](frame-progress.md)
 
@@ -10,7 +10,7 @@ See how a Frame is processed through a Transport, a Pipeline, and a series of Fr
 
 ## [Example Code](examples/)
 
-The repo includes several example apps in the `examples` directory. The docs explain how they work.
+The repository includes several example apps in the `examples` directory. The docs explain how they work.
 
 ## [API Reference](api/)
 

docs/architecture.md

@@ -1,4 +1,4 @@
-# Daily AI SDK Architecture Guide
+# Pipecat architecture guide
 
 ## Frames
 
@@ -10,8 +10,8 @@ Frame processors operate on frames. Every frame processor implements a `process_
 
 ## Pipelines
 
-Pipelines are lists of frame processors that read from a source queue and send the processed frames to a sink queue. A very simple pipeline might chain an LLM frame processor to a text-to-speech frame processor, with a transport's send queue as its sync. Placing LLM message frames on the pipeline's source queue will cause the LLM's response to be spoken. See example #2 for an implementation of this.
+Pipelines are lists of frame processors linked together. Frame processors can push frames upstream or downstream to their peers. A very simple pipeline might chain an LLM frame processor to a text-to-speech frame processor, with a transport as an output.
 
 ## Transports
 
-Transports provide a receive queue, which is input from "the outside world", and a sink queue, which is data that will be sent "to the outside world". The `LocalTransportService` does this with the local camera, mic, display and speaker. The `DailyTransportService` does this with a WebRTC session joined to a Daily.co room.
+Transports provide input and output frame processors to receive or send frames respectively. For example, the `DailyTransport` does this with a WebRTC session joined to a Daily.co room.

dot-env.template

@@ -2,8 +2,16 @@
 ANTHROPIC_API_KEY=...
 
 # Azure
-SPEECH_KEY=...
-SPEECH_REGION=...
+AZURE_SPEECH_REGION=...
+AZURE_SPEECH_API_KEY=...
+
+AZURE_CHATGPT_API_KEY=...
+AZURE_CHATGPT_ENDPOINT=https://...
+AZURE_CHATGPT_MODEL=...
+
+AZURE_DALLE_API_KEY=...
+AZURE_DALLE_ENDPOINT=https://...
+AZURE_DALLE_MODEL=...
 
 # Daily
 DAILY_API_KEY=...
@@ -14,8 +22,10 @@ ELEVENLABS_API_KEY=...
 ELEVENLABS_VOICE_ID=...
 
 # Fal
-FAL_KEY_ID=...
-FAL_KEY_SECRET=...
+FAL_KEY=...
+
+# Fireworks
+FIREWORKS_API_KEY=...
 
 # PlayHT
 PLAY_HT_USER_ID=...

examples/foundational/01-say-one-thing.py

@@ -1,31 +1,36 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
 import aiohttp
-import logging
 import os
-from dailyai.pipeline.frames import EndFrame, TextFrame
-from dailyai.pipeline.pipeline import Pipeline
+import sys
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
+from pipecat.frames.frames import EndFrame, TextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.task import PipelineTask
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
 
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 async def main(room_url):
     async with aiohttp.ClientSession() as session:
         transport = DailyTransport(
-            room_url,
-            None,
-            "Say One Thing",
-            mic_enabled=True,
-        )
+            room_url, None, "Say One Thing", DailyParams(audio_out_enabled=True))
 
         tts = ElevenLabsTTSService(
             aiohttp_session=session,
@@ -33,21 +38,18 @@ async def main(room_url):
             voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
         )
 
-        pipeline = Pipeline([tts])
+        runner = PipelineRunner()
+
+        task = PipelineTask(Pipeline([tts, transport.output()]))
 
         # Register an event handler so we can play the audio when the
         # participant joins.
         @transport.event_handler("on_participant_joined")
-        async def on_participant_joined(transport, participant):
-            if participant["info"]["isLocal"]:
-                return
-
+        async def on_new_participant_joined(transport, participant):
             participant_name = participant["info"]["userName"] or ''
-            await pipeline.queue_frames([TextFrame("Hello there, " + participant_name + "!"), EndFrame()])
-
-        await transport.run(pipeline)
-        del tts
+            await task.queue_frames([TextFrame(f"Hello there, {participant_name}!"), EndFrame()])
 
+        await runner.run(task)
 
 if __name__ == "__main__":
     (url, token) = configure()

examples/foundational/01a-local-audio.py

@@ -0,0 +1,53 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import aiohttp
+import os
+import sys
+
+from pipecat.frames.frames import EndFrame, TextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.transports.base_transport import TransportParams
+from pipecat.transports.local.audio import LocalAudioTransport
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+async def main():
+    async with aiohttp.ClientSession() as session:
+        transport = LocalAudioTransport(TransportParams(audio_out_enabled=True))
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
+        )
+
+        pipeline = Pipeline([tts, transport.output()])
+
+        task = PipelineTask(pipeline)
+
+        async def say_something():
+            await asyncio.sleep(1)
+            await task.queue_frames([TextFrame("Hello there!"), EndFrame()])
+
+        runner = PipelineRunner()
+
+        await asyncio.gather(runner.run(task), say_something())
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/foundational/01a-local-transport.py

@@ -1,38 +0,0 @@
-import asyncio
-import aiohttp
-import logging
-import os
-
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.transports.local_transport import LocalTransport
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-
-async def main():
-    async with aiohttp.ClientSession() as session:
-        meeting_duration_minutes = 1
-        transport = LocalTransport(
-            duration_minutes=meeting_duration_minutes, mic_enabled=True
-        )
-        tts = ElevenLabsTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("ELEVENLABS_API_KEY"),
-            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
-        )
-
-        async def say_something():
-            await asyncio.sleep(1)
-            await transport.say("Hello there.", tts)
-            await transport.stop_when_done()
-
-        await asyncio.gather(transport.run(), say_something())
-
-
-if __name__ == "__main__":
-    asyncio.run(main())

examples/foundational/02-llm-say-one-thing.py

@@ -1,23 +1,31 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
-import os
-import logging
-
 import aiohttp
+import os
+import sys
 
-from dailyai.pipeline.frames import EndFrame, LLMMessagesFrame
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.open_ai_services import OpenAILLMService
+from pipecat.frames.frames import EndFrame, LLMMessagesFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.openai import OpenAILLMService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
 
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 async def main(room_url):
@@ -26,8 +34,7 @@ async def main(room_url):
             room_url,
             None,
             "Say One Thing From an LLM",
-            mic_enabled=True,
-        )
+            DailyParams(audio_out_enabled=True))
 
         tts = ElevenLabsTTSService(
             aiohttp_session=session,
@@ -45,13 +52,15 @@ async def main(room_url):
                 "content": "You are an LLM in a WebRTC session, and this is a 'hello world' demo. Say hello to the world.",
             }]
 
-        pipeline = Pipeline([llm, tts])
+        runner = PipelineRunner()
 
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
-            await pipeline.queue_frames([LLMMessagesFrame(messages), EndFrame()])
+        task = PipelineTask(Pipeline([llm, tts, transport.output()]))
 
-        await transport.run(pipeline)
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            await task.queue_frames([LLMMessagesFrame(messages), EndFrame()])
+
+        await runner.run(task)
 
 
 if __name__ == "__main__":

examples/foundational/03-still-frame.py

@@ -1,21 +1,30 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
 import aiohttp
-import logging
 import os
+import sys
 
-from dailyai.pipeline.frames import TextFrame
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.fal_ai_services import FalImageGenService
+from pipecat.frames.frames import TextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.services.fal import FalImageGenService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
 
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 async def main(room_url):
@@ -24,32 +33,34 @@ async def main(room_url):
             room_url,
             None,
             "Show a still frame image",
-            camera_enabled=True,
-            camera_width=1024,
-            camera_height=1024,
-            duration_minutes=1
+            DailyParams(
+                camera_out_enabled=True,
+                camera_out_width=1024,
+                camera_out_height=1024
+            )
         )
 
         imagegen = FalImageGenService(
-            image_size="square_hd",
+            params=FalImageGenService.InputParams(
+                image_size="square_hd"
+            ),
             aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
+            key=os.getenv("FAL_KEY"),
         )
 
-        pipeline = Pipeline([imagegen])
+        runner = PipelineRunner()
 
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        task = PipelineTask(Pipeline([imagegen, transport.output()]))
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
             # Note that we do not put an EndFrame() item in the pipeline for this demo.
             # This means that the bot will stay in the channel until it times out.
             # An EndFrame() in the pipeline would cause the transport to shut
             # down.
-            await pipeline.queue_frames(
-                [TextFrame("a cat in the style of picasso")]
-            )
+            await task.queue_frames([TextFrame("a cat in the style of picasso")])
 
-        await transport.run(pipeline)
+        await runner.run(task)
 
 
 if __name__ == "__main__":

examples/foundational/03a-image-local.py

@@ -1,57 +0,0 @@
-import asyncio
-import aiohttp
-import logging
-import os
-
-import tkinter as tk
-
-from dailyai.pipeline.frames import TextFrame, EndFrame
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.services.fal_ai_services import FalImageGenService
-from dailyai.transports.local_transport import LocalTransport
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-
-async def main():
-    async with aiohttp.ClientSession() as session:
-        meeting_duration_minutes = 2
-
-        tk_root = tk.Tk()
-        tk_root.title("dailyai")
-
-        transport = LocalTransport(
-            tk_root=tk_root,
-            mic_enabled=False,
-            camera_enabled=True,
-            camera_width=1024,
-            camera_height=1024,
-            duration_minutes=meeting_duration_minutes,
-        )
-
-        imagegen = FalImageGenService(
-            image_size="square_hd",
-            aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
-        )
-
-        pipeline = Pipeline([imagegen])
-        await pipeline.queue_frames([TextFrame("a cat in the style of picasso")])
-
-        async def run_tk():
-            while not transport._stop_threads.is_set():
-                tk_root.update()
-                tk_root.update_idletasks()
-                await asyncio.sleep(0.1)
-
-        await asyncio.gather(transport.run(pipeline, override_pipeline_source_queue=False), run_tk())
-
-
-if __name__ == "__main__":
-    asyncio.run(main())

examples/foundational/03a-local-still-frame.py

@@ -0,0 +1,68 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import aiohttp
+import os
+import sys
+
+import tkinter as tk
+
+from pipecat.frames.frames import TextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.services.fal import FalImageGenService
+from pipecat.transports.base_transport import TransportParams
+from pipecat.transports.local.tk import TkLocalTransport
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+async def main():
+    async with aiohttp.ClientSession() as session:
+        tk_root = tk.Tk()
+        tk_root.title("Picasso Cat")
+
+        transport = TkLocalTransport(
+            tk_root,
+            TransportParams(
+                camera_out_enabled=True,
+                camera_out_width=1024,
+                camera_out_height=1024))
+
+        imagegen = FalImageGenService(
+            params=FalImageGenService.InputParams(
+                image_size="square_hd"
+            ),
+            aiohttp_session=session,
+            key=os.getenv("FAL_KEY"),
+        )
+
+        pipeline = Pipeline([imagegen, transport.output()])
+
+        task = PipelineTask(pipeline)
+        await task.queue_frames([TextFrame("a cat in the style of picasso")])
+
+        runner = PipelineRunner()
+
+        async def run_tk():
+            while runner.is_active():
+                tk_root.update()
+                tk_root.update_idletasks()
+                await asyncio.sleep(0.1)
+
+        await asyncio.gather(runner.run(task), run_tk())
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/foundational/04-utterance-and-speech.py

@@ -1,37 +1,40 @@
-import asyncio
-import logging
-import os
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
 
 import aiohttp
-from dailyai.pipeline.merge_pipeline import SequentialMergePipeline
-from dailyai.pipeline.pipeline import Pipeline
+import asyncio
+import os
+import sys
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
-from dailyai.services.deepgram_ai_services import DeepgramTTSService
-from dailyai.pipeline.frames import EndPipeFrame, LLMMessagesFrame, TextFrame
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
+from pipecat.pipeline.merge_pipeline import SequentialMergePipeline
+from pipecat.pipeline.pipeline import Pipeline
+
+from pipecat.frames.frames import EndPipeFrame, LLMMessagesFrame, TextFrame
+from pipecat.pipeline.task import PipelineTask
+from pipecat.services.azure import AzureLLMService, AzureTTSService
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.transport_services import TransportServiceOutput
+from pipecat.services.transports.daily_transport import DailyTransport
 
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 async def main(room_url: str):
     async with aiohttp.ClientSession() as session:
-        transport = DailyTransport(
-            room_url,
-            None,
-            "Static And Dynamic Speech",
-            duration_minutes=1,
-            mic_enabled=True,
-            mic_sample_rate=16000,
-        )
+        transport = DailyTransport(room_url, None, "Static And Dynamic Speech")
+
+        meeting = TransportServiceOutput(transport, mic_enabled=True)
 
         llm = AzureLLMService(
             api_key=os.getenv("AZURE_CHATGPT_API_KEY"),
@@ -43,10 +46,6 @@ async def main(room_url: str):
             region=os.getenv("AZURE_SPEECH_REGION"),
         )
 
-        deepgram_tts = DeepgramTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("DEEPGRAM_API_KEY"),
-        )
         elevenlabs_tts = ElevenLabsTTSService(
             aiohttp_session=session,
             api_key=os.getenv("ELEVENLABS_API_KEY"),
@@ -56,11 +55,13 @@ async def main(room_url: str):
         messages = [{"role": "system",
                      "content": "tell the user a joke about llamas"}]
 
-        # Start a task to run the LLM to create a joke, and convert the LLM output to audio frames. This task
-        # will run in parallel with generating and speaking the audio for static text, so there's no delay to
-        # speak the LLM response.
+        # Start a task to run the LLM to create a joke, and convert the LLM
+        # output to audio frames. This task will run in parallel with generating
+        # and speaking the audio for static text, so there's no delay to speak
+        # the LLM response.
         llm_pipeline = Pipeline([llm, elevenlabs_tts])
-        await llm_pipeline.queue_frames([LLMMessagesFrame(messages), EndPipeFrame()])
+        llm_task = PipelineTask(llm_pipeline)
+        await llm_task.queue_frames([LLMMessagesFrame(messages), EndPipeFrame()])
 
         simple_tts_pipeline = Pipeline([azure_tts])
         await simple_tts_pipeline.queue_frames(

examples/foundational/05-sync-speech-and-image.py

@@ -1,64 +1,74 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
 import aiohttp
 import os
-import logging
+import sys
 
 from dataclasses import dataclass
-from typing import AsyncGenerator
 
-from dailyai.pipeline.aggregators import (
-    GatedAggregator,
-    LLMFullResponseAggregator,
-    ParallelPipeline,
-    SentenceAggregator,
-)
-from dailyai.pipeline.frames import (
+from pipecat.frames.frames import (
+    AppFrame,
     Frame,
+    ImageRawFrame,
     TextFrame,
     EndFrame,
-    ImageFrame,
     LLMMessagesFrame,
     LLMResponseStartFrame,
 )
-from dailyai.pipeline.frame_processor import FrameProcessor
-
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.fal_ai_services import FalImageGenService
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.processors.aggregators.gated import GatedAggregator
+from pipecat.processors.aggregators.llm_response import LLMFullResponseAggregator
+from pipecat.processors.aggregators.sentence import SentenceAggregator
+from pipecat.processors.aggregators.parallel_task import ParallelTask
+from pipecat.services.openai import OpenAILLMService
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.fal import FalImageGenService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
 
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 @dataclass
-class MonthFrame(Frame):
+class MonthFrame(AppFrame):
     month: str
 
+    def __str__(self):
+        return f"{self.name}(month: {self.month})"
+
 
 class MonthPrepender(FrameProcessor):
     def __init__(self):
+        super().__init__()
         self.most_recent_month = "Placeholder, month frame not yet received"
         self.prepend_to_next_text_frame = False
 
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
         if isinstance(frame, MonthFrame):
             self.most_recent_month = frame.month
         elif self.prepend_to_next_text_frame and isinstance(frame, TextFrame):
-            yield TextFrame(f"{self.most_recent_month}: {frame.text}")
+            await self.push_frame(TextFrame(f"{self.most_recent_month}: {frame.text}"))
             self.prepend_to_next_text_frame = False
         elif isinstance(frame, LLMResponseStartFrame):
             self.prepend_to_next_text_frame = True
-            yield frame
+            await self.push_frame(frame)
         else:
-            yield frame
+            await self.push_frame(frame, direction)
 
 
 async def main(room_url):
@@ -67,11 +77,12 @@ async def main(room_url):
             room_url,
             None,
             "Month Narration Bot",
-            mic_enabled=True,
-            camera_enabled=True,
-            mic_sample_rate=16000,
-            camera_width=1024,
-            camera_height=1024,
+            DailyParams(
+                audio_out_enabled=True,
+                camera_out_enabled=True,
+                camera_out_width=1024,
+                camera_out_height=1024
+            )
         )
 
         tts = ElevenLabsTTSService(
@@ -85,31 +96,33 @@ async def main(room_url):
             model="gpt-4-turbo-preview")
 
         imagegen = FalImageGenService(
-            image_size="square_hd",
+            params=FalImageGenService.InputParams(
+                image_size="square_hd"
+            ),
             aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
+            key=os.getenv("FAL_KEY"),
         )
 
         gated_aggregator = GatedAggregator(
-            gate_open_fn=lambda frame: isinstance(
-                frame, ImageFrame), gate_close_fn=lambda frame: isinstance(
-                frame, LLMResponseStartFrame), start_open=False, )
+            gate_open_fn=lambda frame: isinstance(frame, ImageRawFrame),
+            gate_close_fn=lambda frame: isinstance(frame, LLMResponseStartFrame),
+            start_open=False
+        )
 
         sentence_aggregator = SentenceAggregator()
         month_prepender = MonthPrepender()
         llm_full_response_aggregator = LLMFullResponseAggregator()
 
-        pipeline = Pipeline(
-            processors=[
-                llm,
-                sentence_aggregator,
-                ParallelPipeline(
-                    [[month_prepender, tts], [llm_full_response_aggregator, imagegen]]
-                ),
-                gated_aggregator,
-            ],
-        )
+        pipeline = Pipeline([
+            llm,
+            sentence_aggregator,
+            ParallelTask(
+                [month_prepender, tts],
+                [llm_full_response_aggregator, imagegen]
+            ),
+            gated_aggregator,
+            transport.output()
+        ])
 
         frames = []
         for month in [
@@ -132,13 +145,18 @@ async def main(room_url):
                     "content": f"Describe a nature photograph suitable for use in a calendar, for the month of {month}. Include only the image description with no preamble. Limit the description to one sentence, please.",
                 }
             ]
-            frames.append(MonthFrame(month))
+            frames.append(MonthFrame(month=month))
             frames.append(LLMMessagesFrame(messages))
 
         frames.append(EndFrame())
-        await pipeline.queue_frames(frames)
 
-        await transport.run(pipeline, override_pipeline_source_queue=False)
+        runner = PipelineRunner()
+
+        task = PipelineTask(pipeline)
+
+        await task.queue_frames(frames)
+
+        await runner.run(task)
 
 
 if __name__ == "__main__":

examples/foundational/05a-local-sync-speech-and-image.py

@@ -0,0 +1,164 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import aiohttp
+import asyncio
+import os
+import sys
+
+import tkinter as tk
+
+from pipecat.frames.frames import AudioRawFrame, Frame, URLImageRawFrame, LLMMessagesFrame, TextFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.llm_response import LLMFullResponseAggregator
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.openai import OpenAILLMService
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.fal import FalImageGenService
+from pipecat.transports.base_transport import TransportParams
+from pipecat.transports.local.tk import TkLocalTransport
+
+from loguru import logger
+
+from dotenv import load_dotenv
+
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+async def main():
+    async with aiohttp.ClientSession() as session:
+        tk_root = tk.Tk()
+        tk_root.title("Calendar")
+
+        runner = PipelineRunner()
+
+        async def get_month_data(month):
+            messages = [{"role": "system", "content": f"Describe a nature photograph suitable for use in a calendar, for the month of {month}. Include only the image description with no preamble. Limit the description to one sentence, please.", }]
+
+            class ImageDescription(FrameProcessor):
+                def __init__(self):
+                    super().__init__()
+                    self.text = ""
+
+                async def process_frame(self, frame: Frame, direction: FrameDirection):
+                    if isinstance(frame, TextFrame):
+                        self.text = frame.text
+                    await self.push_frame(frame, direction)
+
+            class AudioGrabber(FrameProcessor):
+                def __init__(self):
+                    super().__init__()
+                    self.audio = bytearray()
+
+                async def process_frame(self, frame: Frame, direction: FrameDirection):
+                    if isinstance(frame, AudioRawFrame):
+                        self.audio.extend(frame.audio)
+                        self.frame = AudioRawFrame(
+                            bytes(self.audio), frame.sample_rate, frame.num_channels)
+
+            class ImageGrabber(FrameProcessor):
+                def __init__(self):
+                    super().__init__()
+                    self.frame = None
+
+                async def process_frame(self, frame: Frame, direction: FrameDirection):
+                    if isinstance(frame, URLImageRawFrame):
+                        self.frame = frame
+
+            llm = OpenAILLMService(
+                api_key=os.getenv("OPENAI_API_KEY"),
+                model="gpt-4-turbo-preview")
+
+            tts = ElevenLabsTTSService(
+                aiohttp_session=session,
+                api_key=os.getenv("ELEVENLABS_API_KEY"),
+                voice_id=os.getenv("ELEVENLABS_VOICE_ID"))
+
+            imagegen = FalImageGenService(
+                params=FalImageGenService.InputParams(
+                    image_size="square_hd"
+                ),
+                aiohttp_session=session,
+                key=os.getenv("FAL_KEY"))
+
+            aggregator = LLMFullResponseAggregator()
+
+            description = ImageDescription()
+
+            audio_grabber = AudioGrabber()
+
+            image_grabber = ImageGrabber()
+
+            pipeline = Pipeline([llm, aggregator, description,
+                                 ParallelPipeline([tts, audio_grabber],
+                                                  [imagegen, image_grabber])])
+
+            task = PipelineTask(pipeline)
+            await task.queue_frame(LLMMessagesFrame(messages))
+            await task.stop_when_done()
+
+            await runner.run(task)
+
+            return {
+                "month": month,
+                "text": description.text,
+                "image": image_grabber.frame,
+                "audio": audio_grabber.frame,
+            }
+
+        transport = TkLocalTransport(
+            tk_root,
+            TransportParams(
+                audio_out_enabled=True,
+                camera_out_enabled=True,
+                camera_out_width=1024,
+                camera_out_height=1024))
+
+        pipeline = Pipeline([transport.output()])
+
+        task = PipelineTask(pipeline)
+
+        # We only specify 5 months as we create tasks all at once and we might
+        # get rate limited otherwise.
+        months: list[str] = [
+            "January",
+            "February",
+            # "March",
+            # "April",
+            # "May",
+        ]
+
+        # We create one task per month. This will be executed concurrently.
+        month_tasks = [asyncio.create_task(get_month_data(month)) for month in months]
+
+        # Now we wait for each month task in the order they're completed. The
+        # benefit is we'll have as little delay as possible before the first
+        # month, and likely no delay between months, but the months won't
+        # display in order.
+        async def show_images(month_tasks):
+            for month_data_task in asyncio.as_completed(month_tasks):
+                data = await month_data_task
+                await task.queue_frames([data["image"], data["audio"]])
+
+            await runner.stop_when_done()
+
+        async def run_tk():
+            while True:
+                tk_root.update()
+                tk_root.update_idletasks()
+                await asyncio.sleep(0.1)
+
+        await asyncio.gather(runner.run(task), show_images(month_tasks), run_tk())
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/foundational/06-listen-and-respond.py

@@ -1,26 +1,37 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
 import aiohttp
-import logging
 import os
-from dailyai.pipeline.frames import LLMMessagesFrame
-from dailyai.pipeline.pipeline import Pipeline
+import sys
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.ai_services import FrameLogger
-from dailyai.pipeline.aggregators import (
-    LLMAssistantContextAggregator,
-    LLMUserContextAggregator,
+from pipecat.frames.frames import LLMMessagesFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.llm_response import (
+    LLMAssistantResponseAggregator,
+    LLMUserResponseAggregator,
 )
+from pipecat.processors.logger import FrameLogger
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.openai import OpenAILLMService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
+from pipecat.vad.silero import SileroVAD
+
 from runner import configure
 
+from loguru import logger
+
 from dotenv import load_dotenv
 load_dotenv(override=True)
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
 async def main(room_url: str, token):
@@ -29,14 +40,15 @@ async def main(room_url: str, token):
             room_url,
             token,
             "Respond bot",
-            duration_minutes=5,
-            start_transcription=True,
-            mic_enabled=True,
-            mic_sample_rate=16000,
-            camera_enabled=False,
-            vad_enabled=True,
+            DailyParams(
+                audio_in_enabled=True,  # This is so Silero VAD can get audio data
+                audio_out_enabled=True,
+                transcription_enabled=True
+            )
         )
 
+        vad = SileroVAD()
+
         tts = ElevenLabsTTSService(
             aiohttp_session=session,
             api_key=os.getenv("ELEVENLABS_API_KEY"),
@@ -46,41 +58,35 @@ async def main(room_url: str, token):
         llm = OpenAILLMService(
             api_key=os.getenv("OPENAI_API_KEY"),
             model="gpt-4-turbo-preview")
-        fl = FrameLogger("Inner")
-        fl2 = FrameLogger("Outer")
+
+        fl_in = FrameLogger("Inner")
+        fl_out = FrameLogger("Outer")
+
         messages = [
             {
                 "role": "system",
-                "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio. Respond to what the user said in a creative and helpful way.",
+                "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio so it should not contain special characters. Respond to what the user said in a creative and helpful way.",
             },
         ]
+        tma_in = LLMUserResponseAggregator(messages)
+        tma_out = LLMAssistantResponseAggregator(messages)
 
-        tma_in = LLMUserContextAggregator(
-            messages, transport._my_participant_id)
-        tma_out = LLMAssistantContextAggregator(
-            messages, transport._my_participant_id
-        )
-        pipeline = Pipeline(
-            processors=[
-                fl,
-                tma_in,
-                llm,
-                fl2,
-                tts,
-                tma_out,
-            ],
-        )
+        pipeline = Pipeline([fl_in, transport.input(), vad, tma_in, llm,
+                            fl_out, tts, tma_out, transport.output()])
 
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        task = PipelineTask(pipeline)
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            transport.capture_participant_transcription(participant["id"])
             # Kick off the conversation.
             messages.append(
                 {"role": "system", "content": "Please introduce yourself to the user."})
-            await pipeline.queue_frames([LLMMessagesFrame(messages)])
+            await task.queue_frames([LLMMessagesFrame(messages)])
 
-        transport.transcription_settings["extra"]["endpointing"] = True
-        transport.transcription_settings["extra"]["punctuate"] = True
-        await transport.run(pipeline)
+        runner = PipelineRunner()
+
+        await runner.run(task)
 
 
 if __name__ == "__main__":

examples/foundational/06a-image-sync.py

@@ -0,0 +1,116 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import aiohttp
+import os
+import sys
+
+from PIL import Image
+
+from pipecat.frames.frames import ImageRawFrame, Frame, SystemFrame, TextFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.llm_context import (
+    LLMAssistantContextAggregator,
+    LLMUserContextAggregator,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.openai import OpenAILLMService
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.transports.services.daily import DailyTransport
+
+from pipecat.transports.services.daily import DailyParams
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+class ImageSyncAggregator(FrameProcessor):
+    def __init__(self, speaking_path: str, waiting_path: str):
+        super().__init__()
+        self._speaking_image = Image.open(speaking_path)
+        self._speaking_image_format = self._speaking_image.format
+        self._speaking_image_bytes = self._speaking_image.tobytes()
+
+        self._waiting_image = Image.open(waiting_path)
+        self._waiting_image_format = self._waiting_image.format
+        self._waiting_image_bytes = self._waiting_image.tobytes()
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if not isinstance(frame, SystemFrame):
+            await self.push_frame(ImageRawFrame(image=self._speaking_image_bytes, size=(1024, 1024), format=self._speaking_image_format))
+            await self.push_frame(frame)
+            await self.push_frame(ImageRawFrame(image=self._waiting_image_bytes, size=(1024, 1024), format=self._waiting_image_format))
+        else:
+            await self.push_frame(frame)
+
+
+async def main(room_url: str, token):
+    async with aiohttp.ClientSession() as session:
+        transport = DailyTransport(
+            room_url,
+            token,
+            "Respond bot",
+            DailyParams(
+                audio_out_enabled=True,
+                camera_out_width=1024,
+                camera_out_height=1024,
+                transcription_enabled=True
+            )
+        )
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
+        )
+
+        llm = OpenAILLMService(
+            api_key=os.getenv("OPENAI_API_KEY"),
+            model="gpt-4-turbo-preview")
+
+        messages = [
+            {
+                "role": "system",
+                "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio so it should not contain special characters. Respond to what the user said in a creative and helpful way.",
+            },
+        ]
+
+        tma_in = LLMUserContextAggregator(messages)
+        tma_out = LLMAssistantContextAggregator(messages)
+
+        image_sync_aggregator = ImageSyncAggregator(
+            os.path.join(os.path.dirname(__file__), "assets", "speaking.png"),
+            os.path.join(os.path.dirname(__file__), "assets", "waiting.png"),
+        )
+
+        pipeline = Pipeline([transport.input(), image_sync_aggregator,
+                            tma_in, llm, tma_out, tts, transport.output()])
+
+        task = PipelineTask(pipeline)
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            participant_name = participant["info"]["userName"] or ''
+            transport.capture_participant_transcription(participant["id"])
+            await task.queue_frames([TextFrame(f"Hi, this is {participant_name}.")])
+
+        runner = PipelineRunner()
+
+        await runner.run(task)
+
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/07-interruptible.py

@@ -2,16 +2,16 @@ import asyncio
 import aiohttp
 import logging
 import os
-from dailyai.pipeline.aggregators import (
-    LLMResponseAggregator,
-    UserResponseAggregator,
+from pipecat.pipeline.aggregators import (
+    LLMAssistantResponseAggregator,
+    LLMUserResponseAggregator,
 )
 
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.services.ai_services import FrameLogger
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.services.ai_services import FrameLogger
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.open_ai_services import OpenAILLMService
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
 
 from runner import configure
 
@@ -19,7 +19,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 
@@ -50,7 +50,7 @@ async def main(room_url: str, token):
         pipeline = Pipeline([FrameLogger(), llm, FrameLogger(), tts])
 
         @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        async def on_first_other_participant_joined(transport, participant):
             await transport.say("Hi, I'm listening!", tts)
 
         async def run_conversation():
@@ -63,11 +63,10 @@ async def main(room_url: str, token):
 
             await transport.run_interruptible_pipeline(
                 pipeline,
-                post_processor=LLMResponseAggregator(messages),
-                pre_processor=UserResponseAggregator(messages),
+                post_processor=LLMAssistantResponseAggregator(messages),
+                pre_processor=LLMUserResponseAggregator(messages),
             )
 
-        transport.transcription_settings["extra"]["punctuate"] = False
         await asyncio.gather(transport.run(), run_conversation())
 
 

examples/foundational/08-bots-arguing.py

@@ -3,14 +3,14 @@ import aiohttp
 import asyncio
 import logging
 import os
-from dailyai.pipeline.aggregators import SentenceAggregator
-from dailyai.pipeline.pipeline import Pipeline
+from pipecat.pipeline.aggregators import SentenceAggregator
+from pipecat.pipeline.pipeline import Pipeline
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.fal_ai_services import FalImageGenService
-from dailyai.pipeline.frames import AudioFrame, EndFrame, ImageFrame, LLMMessagesFrame, TextFrame
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.azure_ai_services import AzureLLMService, AzureTTSService
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
+from pipecat.services.fal_ai_services import FalImageGenService
+from pipecat.pipeline.frames import AudioFrame, EndFrame, ImageFrame, LLMMessagesFrame, TextFrame
 
 from runner import configure
 
@@ -18,7 +18,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 
@@ -51,10 +51,11 @@ async def main(room_url: str):
             voice_id="jBpfuIE2acCO8z3wKNLl",
         )
         dalle = FalImageGenService(
-            image_size="1024x1024",
+            params=FalImageGenService.InputParams(
+                image_size="1024x1024"
+            ),
             aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
+            key=os.getenv("FAL_KEY"),
         )
 
         bot1_messages = [
@@ -91,7 +92,7 @@ async def main(room_url: str):
                 if isinstance(frame, TextFrame):
                     message += frame.text
                 elif isinstance(frame, AudioFrame):
-                    all_audio.extend(frame.data)
+                    all_audio.extend(frame.audio)
 
             return (message, all_audio)
 
@@ -122,7 +123,7 @@ async def main(room_url: str):
                 )
                 await transport.send_queue.put(
                     [
-                        ImageFrame(None, image_data1[1]),
+                        ImageFrame(image_data1[1], image_data1[2]),
                         AudioFrame(audio1),
                     ]
                 )
@@ -134,7 +135,7 @@ async def main(room_url: str):
                 )
                 await transport.send_queue.put(
                     [
-                        ImageFrame(None, image_data2[1]),
+                        ImageFrame(image_data2[1], image_data2[2]),
                         AudioFrame(audio2),
                     ]
                 )

examples/foundational/09-mirror.py

@@ -0,0 +1,62 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import sys
+
+from pipecat.frames.frames import AudioRawFrame, ImageRawFrame
+from pipecat.processors.filter import Filter
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.transports.services.daily import DailyTransport, DailyParams
+
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+async def main(room_url, token):
+    transport = DailyTransport(
+        room_url, token, "Test",
+        DailyParams(
+            audio_in_enabled=True,
+            audio_out_enabled=True,
+            camera_out_enabled=True,
+            camera_out_width=1280,
+            camera_out_height=720
+        )
+    )
+
+    @transport.event_handler("on_first_participant_joined")
+    async def on_first_participant_joined(transport, participant):
+        transport.capture_participant_video(participant["id"])
+
+    # The ParallelPipeline is not really necessary here but it shows how you
+    # would process audio and video concurrently in parallel pipelines.
+    pipeline = Pipeline([transport.input(),
+                         ParallelPipeline(
+                             [Filter([AudioRawFrame])],
+                             [Filter([ImageRawFrame])]),
+                         transport.output()])
+
+    runner = PipelineRunner()
+
+    task = PipelineTask(pipeline)
+
+    await runner.run(task)
+
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/09a-local-mirror.py

@@ -0,0 +1,65 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import sys
+
+import tkinter as tk
+
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.transports.base_transport import TransportParams
+from pipecat.transports.local.tk import TkLocalTransport
+from pipecat.transports.services.daily import DailyParams, DailyTransport
+
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+async def main(room_url, token):
+    tk_root = tk.Tk()
+    tk_root.title("Local Mirror")
+
+    daily_transport = DailyTransport(room_url, token, "Test", DailyParams(audio_in_enabled=True))
+
+    tk_transport = TkLocalTransport(
+        tk_root,
+        TransportParams(
+            audio_out_enabled=True,
+            camera_out_enabled=True,
+            camera_out_width=1280,
+            camera_out_height=720))
+
+    @daily_transport.event_handler("on_first_participant_joined")
+    async def on_first_participant_joined(transport, participant):
+        transport.capture_participant_video(participant["id"])
+
+    pipeline = Pipeline([daily_transport.input(), tk_transport.output()])
+
+    runner = PipelineRunner()
+
+    async def run_tk():
+        while runner.is_active():
+            tk_root.update()
+            tk_root.update_idletasks()
+            await asyncio.sleep(0.1)
+
+    task = PipelineTask(pipeline)
+
+    await asyncio.gather(runner.run(task), run_tk())
+
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/10-wake-word.py

@@ -0,0 +1,181 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import aiohttp
+import os
+import random
+import sys
+
+from PIL import Image
+
+from pipecat.frames.frames import (
+    Frame,
+    SystemFrame,
+    TextFrame,
+    ImageRawFrame,
+    SpriteFrame,
+    TranscriptionFrame,
+)
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.llm_context import (
+    LLMUserContextAggregator,
+    LLMAssistantContextAggregator,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.openai import OpenAILLMService
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
+
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+sprites = {}
+image_files = [
+    "sc-default.png",
+    "sc-talk.png",
+    "sc-listen-1.png",
+    "sc-think-1.png",
+    "sc-think-2.png",
+    "sc-think-3.png",
+    "sc-think-4.png",
+]
+
+script_dir = os.path.dirname(__file__)
+
+for file in image_files:
+    # Build the full path to the image file
+    full_path = os.path.join(script_dir, "assets", file)
+    # Get the filename without the extension to use as the dictionary key
+    filename = os.path.splitext(os.path.basename(full_path))[0]
+    # Open the image and convert it to bytes
+    with Image.open(full_path) as img:
+        sprites[file] = ImageRawFrame(image=img.tobytes(), size=img.size, format=img.format)
+
+# When the bot isn't talking, show a static image of the cat listening
+quiet_frame = sprites["sc-listen-1.png"]
+
+# When the bot is talking, build an animation from two sprites
+talking_list = [sprites["sc-default.png"], sprites["sc-talk.png"]]
+talking = [random.choice(talking_list) for x in range(30)]
+talking_frame = SpriteFrame(talking)
+
+# TODO: Support "thinking" as soon as we get a valid transcript, while LLM
+# is processing
+thinking_list = [
+    sprites["sc-think-1.png"],
+    sprites["sc-think-2.png"],
+    sprites["sc-think-3.png"],
+    sprites["sc-think-4.png"],
+]
+thinking_frame = SpriteFrame(thinking_list)
+
+
+class NameCheckFilter(FrameProcessor):
+    def __init__(self, names: list[str]):
+        super().__init__()
+        self._names = names
+        self._sentence = ""
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, SystemFrame):
+            await self.push_frame(frame, direction)
+            return
+
+        content: str = ""
+
+        # TODO: split up transcription by participant
+        if isinstance(frame, TranscriptionFrame):
+            content = frame.text
+            self._sentence += content
+            if self._sentence.endswith((".", "?", "!")):
+                if any(name in self._sentence for name in self._names):
+                    await self.push_frame(TextFrame(self._sentence))
+                    self._sentence = ""
+                else:
+                    self._sentence = ""
+        else:
+            await self.push_frame(frame, direction)
+
+
+class ImageSyncAggregator(FrameProcessor):
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        await self.push_frame(talking_frame)
+        await self.push_frame(frame)
+        await self.push_frame(quiet_frame)
+
+
+async def main(room_url: str, token):
+    async with aiohttp.ClientSession() as session:
+        transport = DailyTransport(
+            room_url,
+            token,
+            "Santa Cat",
+            DailyParams(
+                audio_out_enabled=True,
+                camera_out_enabled=True,
+                camera_out_width=720,
+                camera_out_height=1280,
+                camera_out_framerate=10,
+                transcription_enabled=True
+            )
+        )
+
+        llm = OpenAILLMService(
+            api_key=os.getenv("OPENAI_API_KEY"),
+            model="gpt-4-turbo-preview")
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id="jBpfuIE2acCO8z3wKNLl",
+        )
+        isa = ImageSyncAggregator()
+
+        messages = [
+            {
+                "role": "system",
+                "content": "You are Santa Cat, a cat that lives in Santa's workshop at the North Pole. You should be clever, and a bit sarcastic. You should also tell jokes every once in a while.  Your responses should only be a few sentences long.",
+            },
+        ]
+
+        tma_in = LLMUserContextAggregator(messages)
+        tma_out = LLMAssistantContextAggregator(messages)
+        ncf = NameCheckFilter(["Santa Cat", "Santa"])
+
+        pipeline = Pipeline([transport.input(), isa, ncf, tma_in,
+                            llm, tma_out, tts, transport.output()])
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            # Send some greeting at the beginning.
+            await tts.say("Hi! If you want to talk to me, just say 'hey Santa Cat'.")
+            transport.capture_participant_transcription(participant["id"])
+
+        async def starting_image():
+            await transport.send_image(quiet_frame)
+
+        runner = PipelineRunner()
+
+        task = PipelineTask(pipeline)
+
+        await asyncio.gather(runner.run(task), starting_image())
+
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/11-sound-effects.py

@@ -0,0 +1,132 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import aiohttp
+import asyncio
+import os
+import sys
+import wave
+
+from pipecat.frames.frames import (
+    Frame,
+    AudioRawFrame,
+    LLMResponseEndFrame,
+    LLMMessagesFrame,
+)
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.llm_context import (
+    LLMUserContextAggregator,
+    LLMAssistantContextAggregator,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.processors.logger import FrameLogger
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.openai import OpenAILLMService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
+
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+sounds = {}
+sound_files = ["ding1.wav", "ding2.wav"]
+
+script_dir = os.path.dirname(__file__)
+
+for file in sound_files:
+    # Build the full path to the image file
+    full_path = os.path.join(script_dir, "assets", file)
+    # Get the filename without the extension to use as the dictionary key
+    filename = os.path.splitext(os.path.basename(full_path))[0]
+    # Open the image and convert it to bytes
+    with wave.open(full_path) as audio_file:
+        sounds[file] = AudioRawFrame(audio_file.readframes(-1),
+                                     audio_file.getframerate(), audio_file.getnchannels())
+
+
+class OutboundSoundEffectWrapper(FrameProcessor):
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, LLMResponseEndFrame):
+            await self.push_frame(sounds["ding1.wav"])
+            # In case anything else downstream needs it
+            await self.push_frame(frame, direction)
+        else:
+            await self.push_frame(frame, direction)
+
+
+class InboundSoundEffectWrapper(FrameProcessor):
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, LLMMessagesFrame):
+            await self.push_frame(sounds["ding2.wav"])
+            # In case anything else downstream needs it
+            await self.push_frame(frame, direction)
+        else:
+            await self.push_frame(frame, direction)
+
+
+async def main(room_url: str, token):
+    async with aiohttp.ClientSession() as session:
+        transport = DailyTransport(
+            room_url,
+            token,
+            "Respond bot",
+            DailyParams(audio_out_enabled=True, transcription_enabled=True)
+        )
+
+        llm = OpenAILLMService(
+            api_key=os.getenv("OPENAI_API_KEY"),
+            model="gpt-4-turbo-preview")
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id="ErXwobaYiN019PkySvjV",
+        )
+
+        messages = [
+            {
+                "role": "system",
+                "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio. Respond to what the user said in a creative and helpful way.",
+            },
+        ]
+
+        tma_in = LLMUserContextAggregator(messages)
+        tma_out = LLMAssistantContextAggregator(messages)
+        out_sound = OutboundSoundEffectWrapper()
+        in_sound = InboundSoundEffectWrapper()
+        fl = FrameLogger("LLM Out")
+        fl2 = FrameLogger("Transcription In")
+
+        pipeline = Pipeline([transport.input(), tma_in, in_sound, fl2, llm,
+                            tma_out, fl, tts, out_sound, transport.output()])
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            transport.capture_participant_transcription(participant["id"])
+            await tts.say("Hi, I'm listening!")
+            await transport.send_audio(sounds["ding1.wav"])
+
+        runner = PipelineRunner()
+
+        task = PipelineTask(pipeline)
+
+        await runner.run(task)
+
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/12-describe-video.py

@@ -0,0 +1,104 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import aiohttp
+import os
+import sys
+
+from pipecat.frames.frames import Frame, TextFrame, UserImageRequestFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.aggregators.user_response import UserResponseAggregator
+from pipecat.processors.aggregators.vision_image_frame import VisionImageFrameAggregator
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.elevenlabs import ElevenLabsTTSService
+from pipecat.services.moondream import MoondreamService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
+from pipecat.vad.silero import SileroVAD
+
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+class UserImageRequester(FrameProcessor):
+
+    def __init__(self, participant_id: str | None = None):
+        super().__init__()
+        self._participant_id = participant_id
+
+    def set_participant_id(self, participant_id: str):
+        self._participant_id = participant_id
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if self._participant_id and isinstance(frame, TextFrame):
+            await self.push_frame(UserImageRequestFrame(self._participant_id), FrameDirection.UPSTREAM)
+        await self.push_frame(frame, direction)
+
+
+async def main(room_url: str, token):
+    async with aiohttp.ClientSession() as session:
+        transport = DailyTransport(
+            room_url,
+            token,
+            "Describe participant video",
+            DailyParams(
+                audio_in_enabled=True,  # This is so Silero VAD can get audio data
+                audio_out_enabled=True,
+                transcription_enabled=True
+            )
+        )
+
+        vad = SileroVAD()
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
+        )
+
+        user_response = UserResponseAggregator()
+
+        image_requester = UserImageRequester()
+
+        vision_aggregator = VisionImageFrameAggregator()
+
+        # If you run into weird description, try with use_cpu=True
+        moondream = MoondreamService()
+
+        tts = ElevenLabsTTSService(
+            aiohttp_session=session,
+            api_key=os.getenv("ELEVENLABS_API_KEY"),
+            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
+        )
+
+        @transport.event_handler("on_first_participant_joined")
+        async def on_first_participant_joined(transport, participant):
+            await tts.say("Hi there! Feel free to ask me what I see.")
+            transport.capture_participant_video(participant["id"], framerate=0)
+            transport.capture_participant_transcription(participant["id"])
+            image_requester.set_participant_id(participant["id"])
+
+        pipeline = Pipeline([transport.input(), vad, user_response, image_requester,
+                             vision_aggregator, moondream, tts, transport.output()])
+
+        task = PipelineTask(pipeline)
+
+        runner = PipelineRunner()
+
+        await runner.run(task)
+
+if __name__ == "__main__":
+    (url, token) = configure()
+    asyncio.run(main(url, token))

examples/foundational/13-whisper-transcription.py

@@ -1,42 +1,53 @@
-import asyncio
-import logging
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.whisper_ai_services import WhisperSTTService
-from dailyai.pipeline.pipeline import Pipeline
+import asyncio
+import sys
+
+from pipecat.frames.frames import Frame, TranscriptionFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.whisper import WhisperSTTService
+from pipecat.transports.services.daily import DailyParams, DailyTransport
 
 from runner import configure
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+class TranscriptionLogger(FrameProcessor):
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TranscriptionFrame):
+            print(f"Transcription: {frame.text}")
 
 
 async def main(room_url: str):
-    transport = DailyTransport(
-        room_url,
-        None,
-        "Transcription bot",
-        start_transcription=False,
-        mic_enabled=False,
-        camera_enabled=False,
-        speaker_enabled=True,
-    )
+    transport = DailyTransport(room_url, None, "Transcription bot",
+                               DailyParams(audio_in_enabled=True))
 
     stt = WhisperSTTService()
 
-    transcription_output_queue = asyncio.Queue()
+    tl = TranscriptionLogger()
 
-    pipeline = Pipeline([stt])
-    pipeline.set_sink(transcription_output_queue)
+    pipeline = Pipeline([transport.input(), stt, tl])
 
-    async def handle_transcription():
-        print("`````````TRANSCRIPTION`````````")
-        while True:
-            item = await transcription_output_queue.get()
-            print(item.text)
+    task = PipelineTask(pipeline)
 
-    await asyncio.gather(transport.run(pipeline), handle_transcription())
+    runner = PipelineRunner()
+
+    await runner.run(task)
 
 
 if __name__ == "__main__":

examples/foundational/13a-whisper-local.py

@@ -1,53 +1,55 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
 import asyncio
-import logging
+import sys
 
-from dailyai.pipeline.frames import EndFrame, TranscriptionFrame
-from dailyai.transports.local_transport import LocalTransport
-from dailyai.services.whisper_ai_services import WhisperSTTService
-from dailyai.pipeline.pipeline import Pipeline
+from pipecat.frames.frames import Frame, TranscriptionFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineTask
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.services.whisper import WhisperSTTService
+from pipecat.transports.base_transport import TransportParams
+from pipecat.transports.local.audio import LocalAudioTransport
 
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
+from runner import configure
+
+from loguru import logger
+
+from dotenv import load_dotenv
+load_dotenv(override=True)
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
 
 
-async def main():
-    meeting_duration_minutes = 1
+class TranscriptionLogger(FrameProcessor):
 
-    transport = LocalTransport(
-        mic_enabled=False,
-        camera_enabled=False,
-        speaker_enabled=True,
-        duration_minutes=meeting_duration_minutes,
-        start_transcription=False,
-    )
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TranscriptionFrame):
+            print(f"Transcription: {frame.text}")
+
+
+async def main(room_url: str):
+    transport = LocalAudioTransport(TransportParams(audio_in_enabled=True))
 
     stt = WhisperSTTService()
 
-    transcription_output_queue = asyncio.Queue()
-    transport_done = asyncio.Event()
+    tl = TranscriptionLogger()
 
-    pipeline = Pipeline([stt])
-    pipeline.set_sink(transcription_output_queue)
+    pipeline = Pipeline([transport.input(), stt, tl])
 
-    async def handle_transcription():
-        print("`````````TRANSCRIPTION`````````")
-        while not transport_done.is_set():
-            item = await transcription_output_queue.get()
-            print("got item from queue", item)
-            if isinstance(item, TranscriptionFrame):
-                print(item.text)
-            elif isinstance(item, EndFrame):
-                break
-        print("handle_transcription done")
+    task = PipelineTask(pipeline)
 
-    async def run_until_done():
-        await transport.run(pipeline)
-        transport_done.set()
-        print("run_until_done done")
+    runner = PipelineRunner()
 
-    await asyncio.gather(run_until_done(), handle_transcription())
+    await runner.run(task)
 
 
 if __name__ == "__main__":
-    asyncio.run(main())
+    (url, token) = configure()
+    asyncio.run(main(url))

examples/foundational/to_be_updated/05a-local-sync-speech-and-text.py

@@ -1,134 +0,0 @@
-import aiohttp
-import asyncio
-import logging
-import tkinter as tk
-import os
-
-from dailyai.pipeline.frames import AudioFrame, ImageFrame
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.fal_ai_services import FalImageGenService
-from dailyai.transports.local_transport import LocalTransport
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-
-async def main():
-    async with aiohttp.ClientSession() as session:
-        meeting_duration_minutes = 5
-        tk_root = tk.Tk()
-        tk_root.title("Calendar")
-
-        transport = LocalTransport(
-            mic_enabled=True,
-            camera_enabled=True,
-            camera_width=1024,
-            camera_height=1024,
-            duration_minutes=meeting_duration_minutes,
-            tk_root=tk_root,
-        )
-
-        tts = ElevenLabsTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("ELEVENLABS_API_KEY"),
-            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
-        )
-
-        llm = OpenAILLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
-            model="gpt-4-turbo-preview")
-
-        dalle = FalImageGenService(
-            image_size="1024x1024",
-            aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
-        )
-
-        # Get a complete audio chunk from the given text. Splitting this into its own
-        # coroutine lets us ensure proper ordering of the audio chunks on the
-        # send queue.
-        async def get_all_audio(text):
-            all_audio = bytearray()
-            async for audio in tts.run_tts(text):
-                all_audio.extend(audio)
-
-            return all_audio
-
-        async def get_month_data(month):
-            messages = [{"role": "system", "content": f"Describe a nature photograph suitable for use in a calendar, for the month of {
-                month}. Include only the image description with no preamble. Limit the description to one sentence, please.", }]
-
-            image_description = await llm.run_llm(messages)
-            if not image_description:
-                return
-
-            to_speak = f"{month}: {image_description}"
-            audio_task = asyncio.create_task(get_all_audio(to_speak))
-            image_task = asyncio.create_task(
-                dalle.run_image_gen(image_description))
-            (audio, image_data) = await asyncio.gather(audio_task, image_task)
-
-            return {
-                "month": month,
-                "text": image_description,
-                "image_url": image_data[0],
-                "image": image_data[1],
-                "audio": audio,
-            }
-
-        months: list[str] = [
-            "January",
-            "February",
-            "March",
-            "April",
-            "May",
-            "June",
-            "July",
-            "August",
-            "September",
-            "October",
-            "November",
-            "December",
-        ]
-
-        async def show_images():
-            # This will play the months in the order they're completed. The benefit
-            # is we'll have as little delay as possible before the first month, and
-            # likely no delay between months, but the months won't display in
-            # order.
-            for month_data_task in asyncio.as_completed(month_tasks):
-                data = await month_data_task
-                if data:
-                    await transport.send_queue.put(
-                        [
-                            ImageFrame(data["image_url"], data["image"]),
-                            AudioFrame(data["audio"]),
-                        ]
-                    )
-
-            await asyncio.sleep(25)
-
-            # wait for the output queue to be empty, then leave the meeting
-            await transport.stop_when_done()
-
-        async def run_tk():
-            while not transport._stop_threads.is_set():
-                tk_root.update()
-                tk_root.update_idletasks()
-                await asyncio.sleep(0.1)
-
-        month_tasks = [
-            asyncio.create_task(
-                get_month_data(month)) for month in months]
-
-        await asyncio.gather(transport.run(), show_images(), run_tk())
-
-
-if __name__ == "__main__":
-    asyncio.run(main())

examples/foundational/to_be_updated/06a-image-sync.py

@@ -1,122 +0,0 @@
-import asyncio
-import os
-import logging
-from typing import AsyncGenerator
-import aiohttp
-from PIL import Image
-
-from dailyai.pipeline.frames import ImageFrame, Frame
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.ai_services import AIService
-from dailyai.pipeline.aggregators import (
-    LLMAssistantContextAggregator,
-    LLMUserContextAggregator,
-)
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.services.fal_ai_services import FalImageGenService
-
-from runner import configure
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-
-class ImageSyncAggregator(AIService):
-    def __init__(self, speaking_path: str, waiting_path: str):
-        self._speaking_image = Image.open(speaking_path)
-        self._speaking_image_bytes = self._speaking_image.tobytes()
-
-        self._waiting_image = Image.open(waiting_path)
-        self._waiting_image_bytes = self._waiting_image.tobytes()
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        yield ImageFrame(None, self._speaking_image_bytes)
-        yield frame
-        yield ImageFrame(None, self._waiting_image_bytes)
-
-
-async def main(room_url: str, token):
-    async with aiohttp.ClientSession() as session:
-        transport = DailyTransport(
-            room_url,
-            token,
-            "Respond bot",
-            5,
-        )
-        transport._camera_enabled = True
-        transport._camera_width = 1024
-        transport._camera_height = 1024
-        transport._mic_enabled = True
-        transport._mic_sample_rate = 16000
-
-        tts = ElevenLabsTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("ELEVENLABS_API_KEY"),
-            voice_id=os.getenv("ELEVENLABS_VOICE_ID"),
-        )
-
-        llm = OpenAILLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
-            model="gpt-4-turbo-preview")
-
-        img = FalImageGenService(
-            image_size="1024x1024",
-            aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
-        )
-
-        async def get_images():
-            get_speaking_task = asyncio.create_task(
-                img.run_image_gen("An image of a cat speaking")
-            )
-            get_waiting_task = asyncio.create_task(
-                img.run_image_gen("An image of a cat waiting")
-            )
-
-            (speaking_data, waiting_data) = await asyncio.gather(
-                get_speaking_task, get_waiting_task
-            )
-
-            return speaking_data, waiting_data
-
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
-            await tts.say("Hi, I'm listening!", transport.send_queue)
-
-        async def handle_transcriptions():
-            messages = [
-                {
-                    "role": "system",
-                    "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio. Respond to what the user said in a creative and helpful way.",
-                },
-            ]
-
-            tma_in = LLMUserContextAggregator(
-                messages, transport._my_participant_id)
-            tma_out = LLMAssistantContextAggregator(
-                messages, transport._my_participant_id
-            )
-            image_sync_aggregator = ImageSyncAggregator(
-                os.path.join(
-                    os.path.dirname(__file__), "assets", "speaking.png"), os.path.join(
-                    os.path.dirname(__file__), "assets", "waiting.png"), )
-            await tts.run_to_queue(
-                transport.send_queue,
-                image_sync_aggregator.run(
-                    tma_out.run(llm.run(tma_in.run(transport.get_receive_frames())))
-                ),
-            )
-
-        transport.transcription_settings["extra"]["punctuate"] = True
-        await asyncio.gather(transport.run(), handle_transcriptions())
-
-
-if __name__ == "__main__":
-    (url, token) = configure()
-    asyncio.run(main(url, token))

examples/foundational/to_be_updated/10-wake-word.py

@@ -1,191 +0,0 @@
-import aiohttp
-import asyncio
-import logging
-import os
-import random
-from typing import AsyncGenerator
-from PIL import Image
-
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.pipeline.aggregators import (
-    LLMUserContextAggregator,
-    LLMAssistantContextAggregator,
-)
-from dailyai.pipeline.frames import (
-    Frame,
-    TextFrame,
-    ImageFrame,
-    SpriteFrame,
-    TranscriptionFrame,
-)
-from dailyai.services.ai_services import AIService
-
-from runner import configure
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-sprites = {}
-image_files = [
-    "sc-default.png",
-    "sc-talk.png",
-    "sc-listen-1.png",
-    "sc-think-1.png",
-    "sc-think-2.png",
-    "sc-think-3.png",
-    "sc-think-4.png",
-]
-
-script_dir = os.path.dirname(__file__)
-
-for file in image_files:
-    # Build the full path to the image file
-    full_path = os.path.join(script_dir, "assets", file)
-    # Get the filename without the extension to use as the dictionary key
-    filename = os.path.splitext(os.path.basename(full_path))[0]
-    # Open the image and convert it to bytes
-    with Image.open(full_path) as img:
-        sprites[file] = img.tobytes()
-
-# When the bot isn't talking, show a static image of the cat listening
-quiet_frame = ImageFrame("", sprites["sc-listen-1.png"])
-# When the bot is talking, build an animation from two sprites
-talking_list = [sprites["sc-default.png"], sprites["sc-talk.png"]]
-talking = [random.choice(talking_list) for x in range(30)]
-talking_frame = SpriteFrame(images=talking)
-
-# TODO: Support "thinking" as soon as we get a valid transcript, while LLM
-# is processing
-thinking_list = [
-    sprites["sc-think-1.png"],
-    sprites["sc-think-2.png"],
-    sprites["sc-think-3.png"],
-    sprites["sc-think-4.png"],
-]
-thinking_frame = SpriteFrame(images=thinking_list)
-
-
-class TranscriptFilter(AIService):
-    def __init__(self, bot_participant_id=None):
-        self.bot_participant_id = bot_participant_id
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, TranscriptionFrame):
-            if frame.participantId != self.bot_participant_id:
-                yield frame
-
-
-class NameCheckFilter(AIService):
-    def __init__(self, names: list[str]):
-        self.names = names
-        self.sentence = ""
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        content: str = ""
-
-        # TODO: split up transcription by participant
-        if isinstance(frame, TextFrame):
-            content = frame.text
-
-        self.sentence += content
-        if self.sentence.endswith((".", "?", "!")):
-            if any(name in self.sentence for name in self.names):
-                out = self.sentence
-                self.sentence = ""
-                yield TextFrame(out)
-            else:
-                out = self.sentence
-                self.sentence = ""
-
-
-class ImageSyncAggregator(AIService):
-    def __init__(self):
-        pass
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        yield talking_frame
-        yield frame
-        yield quiet_frame
-
-
-async def main(room_url: str, token):
-    async with aiohttp.ClientSession() as session:
-        transport = DailyTransport(
-            room_url,
-            token,
-            "Santa Cat",
-            duration_minutes=3,
-            start_transcription=True,
-            mic_enabled=True,
-            mic_sample_rate=16000,
-            camera_enabled=True,
-            camera_width=720,
-            camera_height=1280,
-        )
-        transport._mic_enabled = True
-        transport._mic_sample_rate = 16000
-        transport._camera_enabled = True
-        transport._camera_width = 720
-        transport._camera_height = 1280
-
-        llm = OpenAILLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
-            model="gpt-4-turbo-preview")
-
-        tts = ElevenLabsTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("ELEVENLABS_API_KEY"),
-            voice_id="jBpfuIE2acCO8z3wKNLl",
-        )
-        isa = ImageSyncAggregator()
-
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
-            await tts.say(
-                "Hi! If you want to talk to me, just say 'hey Santa Cat'.",
-                transport.send_queue,
-            )
-
-        async def handle_transcriptions():
-            messages = [
-                {
-                    "role": "system",
-                    "content": "You are Santa Cat, a cat that lives in Santa's workshop at the North Pole. You should be clever, and a bit sarcastic. You should also tell jokes every once in a while.  Your responses should only be a few sentences long.",
-                },
-            ]
-
-            tma_in = LLMUserContextAggregator(
-                messages, transport._my_participant_id)
-            tma_out = LLMAssistantContextAggregator(
-                messages, transport._my_participant_id
-            )
-            tf = TranscriptFilter(transport._my_participant_id)
-            ncf = NameCheckFilter(["Santa Cat", "Santa"])
-            await tts.run_to_queue(
-                transport.send_queue,
-                isa.run(
-                    tma_out.run(
-                        llm.run(
-                            tma_in.run(
-                                ncf.run(tf.run(transport.get_receive_frames())))
-                        )
-                    )
-                ),
-            )
-
-        async def starting_image():
-            await transport.send_queue.put(quiet_frame)
-
-        transport.transcription_settings["extra"]["punctuate"] = True
-        await asyncio.gather(transport.run(), handle_transcriptions(), starting_image())
-
-
-if __name__ == "__main__":
-    (url, token) = configure()
-    asyncio.run(main(url, token))

examples/foundational/to_be_updated/11-sound-effects.py

@@ -1,140 +0,0 @@
-import aiohttp
-import asyncio
-import logging
-import os
-import wave
-
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.pipeline.aggregators import (
-    LLMUserContextAggregator,
-    LLMAssistantContextAggregator,
-)
-from dailyai.services.ai_services import AIService, FrameLogger
-from dailyai.pipeline.frames import (
-    Frame,
-    AudioFrame,
-    LLMResponseEndFrame,
-    LLMMessagesFrame,
-)
-from typing import AsyncGenerator
-
-from runner import configure
-
-from dotenv import load_dotenv
-load_dotenv(override=True)
-
-logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
-logger.setLevel(logging.DEBUG)
-
-sounds = {}
-sound_files = ["ding1.wav", "ding2.wav"]
-
-script_dir = os.path.dirname(__file__)
-
-for file in sound_files:
-    # Build the full path to the image file
-    full_path = os.path.join(script_dir, "assets", file)
-    # Get the filename without the extension to use as the dictionary key
-    filename = os.path.splitext(os.path.basename(full_path))[0]
-    # Open the image and convert it to bytes
-    with wave.open(full_path) as audio_file:
-        sounds[file] = audio_file.readframes(-1)
-
-
-class OutboundSoundEffectWrapper(AIService):
-    def __init__(self):
-        pass
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, LLMResponseEndFrame):
-            yield AudioFrame(sounds["ding1.wav"])
-            # In case anything else up the stack needs it
-            yield frame
-        else:
-            yield frame
-
-
-class InboundSoundEffectWrapper(AIService):
-    def __init__(self):
-        pass
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, LLMMessagesFrame):
-            yield AudioFrame(sounds["ding2.wav"])
-            # In case anything else up the stack needs it
-            yield frame
-        else:
-            yield frame
-
-
-async def main(room_url: str, token):
-    async with aiohttp.ClientSession() as session:
-        transport = DailyTransport(
-            room_url,
-            token,
-            "Respond bot",
-            duration_minutes=5,
-            mic_enabled=True,
-            mic_sample_rate=16000,
-            camera_enabled=False,
-        )
-
-        llm = OpenAILLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
-            model="gpt-4-turbo-preview")
-
-        tts = ElevenLabsTTSService(
-            aiohttp_session=session,
-            api_key=os.getenv("ELEVENLABS_API_KEY"),
-            voice_id="ErXwobaYiN019PkySvjV",
-        )
-
-        @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
-            await tts.say("Hi, I'm listening!", transport.send_queue)
-            await transport.send_queue.put(AudioFrame(sounds["ding1.wav"]))
-
-        async def handle_transcriptions():
-            messages = [
-                {
-                    "role": "system",
-                    "content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be converted to audio. Respond to what the user said in a creative and helpful way.",
-                },
-            ]
-
-            tma_in = LLMUserContextAggregator(
-                messages, transport._my_participant_id)
-            tma_out = LLMAssistantContextAggregator(
-                messages, transport._my_participant_id
-            )
-            out_sound = OutboundSoundEffectWrapper()
-            in_sound = InboundSoundEffectWrapper()
-            fl = FrameLogger("LLM Out")
-            fl2 = FrameLogger("Transcription In")
-            await out_sound.run_to_queue(
-                transport.send_queue,
-                tts.run(
-                    fl.run(
-                        tma_out.run(
-                            llm.run(
-                                fl2.run(
-                                    in_sound.run(
-                                        tma_in.run(transport.get_receive_frames())
-                                    )
-                                )
-                            )
-                        )
-                    )
-                ),
-            )
-
-        transport.transcription_settings["extra"]["punctuate"] = True
-        await asyncio.gather(transport.run(), handle_transcriptions())
-
-
-if __name__ == "__main__":
-    (url, token) = configure()
-    asyncio.run(main(url, token))

examples/foundational/websocket-server/frames.proto

@@ -1,6 +1,6 @@
 syntax = "proto3";
 
-package dailyai_proto;
+package pipecat_proto;
 
 message TextFrame {
     string text = 1;

examples/foundational/websocket-server/index.html

@@ -28,7 +28,7 @@
 
         const proto = protobuf.load("frames.proto", (err, root) => {
             if (err) throw err;
-            frame = root.lookupType("dailyai_proto.Frame");
+            frame = root.lookupType("pipecat_proto.Frame");
         });
 
         function initWebSocket() {

examples/foundational/websocket-server/sample.py

@@ -2,15 +2,15 @@ import asyncio
 import aiohttp
 import logging
 import os
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.pipeline.frames import TextFrame, TranscriptionFrame
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.transports.websocket_transport import WebsocketTransport
-from dailyai.services.whisper_ai_services import WhisperSTTService
+from pipecat.pipeline.frame_processor import FrameProcessor
+from pipecat.pipeline.frames import TextFrame, TranscriptionFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
+from pipecat.transports.websocket_transport import WebsocketTransport
+from pipecat.services.whisper_ai_services import WhisperSTTService
 
 logging.basicConfig(format="%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 

examples/image-gen.py

@@ -5,11 +5,10 @@ import time
 import urllib.parse
 import random
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
-from dailyai.pipeline.frames import Frame, FrameType
-from dailyai.services.fal_ai_services import FalImageGenService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.azure_ai_services import AzureLLMService, AzureTTSService
+from pipecat.pipeline.frames import Frame
+from pipecat.services.fal_ai_services import FalImageGenService
 
 
 async def main(room_url: str, token):
@@ -77,8 +76,6 @@ async def main(room_url: str, token):
             async for audio in audio_generator:
                 transport.output_queue.put(Frame(FrameType.AUDIO_FRAME, audio))
 
-    transport.transcription_settings["extra"]["punctuate"] = False
-    transport.transcription_settings["extra"]["endpointing"] = False
     await asyncio.gather(transport.run(), handle_transcriptions())
 
 

examples/internal/11a-dial-out.py

@@ -3,11 +3,11 @@ import asyncio
 import os
 import wave
 
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
-from dailyai.pipeline.aggregators import LLMContextAggregator
-from dailyai.services.ai_services import AIService, FrameLogger
-from dailyai.pipeline.frames import Frame, AudioFrame, LLMResponseEndFrame, LLMMessagesFrame
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.azure_ai_services import AzureLLMService, AzureTTSService
+from pipecat.pipeline.aggregators import LLMContextAggregator
+from pipecat.services.ai_services import AIService, FrameLogger
+from pipecat.pipeline.frames import Frame, AudioFrame, LLMResponseEndFrame, LLMMessagesFrame
 from typing import AsyncGenerator
 
 from runner import configure
@@ -80,7 +80,7 @@ async def main(room_url: str, token, phone):
         tts = AzureTTSService()
 
         @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        async def on_first_other_participant_joined(transport, participant):
             await tts.say("Hi, I'm listening!", transport.send_queue)
             await transport.send_queue.put(AudioFrame(sounds["ding1.wav"]))
 
@@ -127,8 +127,6 @@ async def main(room_url: str, token, phone):
                     transport.start_recording()
                     transport.dialout(phone)
 
-        transport.transcription_settings["extra"]["punctuate"] = True
-
         await asyncio.gather(transport.run(), handle_transcriptions())
 
 

examples/starter-apps/chatbot.py

@@ -5,11 +5,11 @@ import os
 from PIL import Image
 from typing import AsyncGenerator
 
-from dailyai.pipeline.aggregators import (
-    LLMResponseAggregator,
-    UserResponseAggregator,
+from pipecat.pipeline.aggregators import (
+    LLMAssistantResponseAggregator,
+    LLMUserResponseAggregator,
 )
-from dailyai.pipeline.frames import (
+from pipecat.pipeline.frames import (
     ImageFrame,
     SpriteFrame,
     Frame,
@@ -18,11 +18,11 @@ from dailyai.pipeline.frames import (
     AudioFrame,
     PipelineStartedFrame,
 )
-from dailyai.services.ai_services import AIService
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
+from pipecat.services.ai_services import AIService
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.open_ai_services import OpenAILLMService
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
 
 from runner import configure
 
@@ -30,7 +30,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 sprites = []
@@ -48,7 +48,7 @@ for i in range(1, 26):
 flipped = sprites[::-1]
 sprites.extend(flipped)
 # When the bot isn't talking, show a static image of the cat listening
-quiet_frame = ImageFrame("", sprites[0])
+quiet_frame = ImageFrame(sprites[0], (1024, 576))
 talking_frame = SpriteFrame(images=sprites)
 
 
@@ -79,8 +79,6 @@ class TalkingAnimation(AIService):
 
 
 class AnimationInitializer(AIService):
-    def __init__(self):
-        super().__init__()
 
     async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
         if isinstance(frame, PipelineStartedFrame):
@@ -127,7 +125,7 @@ async def main(room_url: str, token):
         ]
 
         @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        async def on_first_other_participant_joined(transport, participant):
             print(f"!!! in here, pipeline.source is {pipeline.source}")
             await pipeline.queue_frames([LLMMessagesFrame(messages)])
 
@@ -135,12 +133,10 @@ async def main(room_url: str, token):
 
             await transport.run_interruptible_pipeline(
                 pipeline,
-                post_processor=LLMResponseAggregator(messages),
-                pre_processor=UserResponseAggregator(messages),
+                post_processor=LLMAssistantResponseAggregator(messages),
+                pre_processor=LLMUserResponseAggregator(messages),
             )
 
-        transport.transcription_settings["extra"]["endpointing"] = True
-        transport.transcription_settings["extra"]["punctuate"] = True
         await asyncio.gather(transport.run(), run_conversation())
 
 

examples/starter-apps/patient-intake.py

@@ -7,25 +7,26 @@ import os
 import re
 import wave
 from typing import AsyncGenerator, List
-from dailyai.pipeline.opeanai_llm_aggregator import (
+from pipecat.pipeline.opeanai_llm_aggregator import (
     OpenAIAssistantContextAggregator,
     OpenAIUserContextAggregator,
 )
 
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.openai_llm_context import OpenAILLMContext
-from dailyai.services.open_ai_services import OpenAILLMService
-# from dailyai.services.deepgram_ai_services import DeepgramTTSService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.pipeline.frames import (
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.openai_llm_context import OpenAILLMContext
+from pipecat.services.open_ai_services import OpenAILLMService
+# from pipecat.services.deepgram_ai_services import DeepgramTTSService
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
+from pipecat.services.fireworks_ai_services import FireworksLLMService
+from pipecat.pipeline.frames import (
     Frame,
     LLMFunctionCallFrame,
     LLMFunctionStartFrame,
     AudioFrame,
 )
-from dailyai.pipeline.openai_frames import OpenAILLMContextFrame
-from dailyai.services.ai_services import FrameLogger, AIService
+from pipecat.pipeline.openai_frames import OpenAILLMContextFrame
+from pipecat.services.ai_services import FrameLogger, AIService
 from openai._types import NotGiven, NOT_GIVEN
 
 from openai.types.chat import (
@@ -38,7 +39,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format="%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 sounds = {}
@@ -249,8 +250,7 @@ class ChecklistProcessor(AIService):
                 print(f"--> {pretty_json}\n")
                 if frame.function_name not in self._functions:
                     raise Exception(
-                        f"The LLM tried to call a function named {frame.function_name}, which isn't in the list of known functions. Please check your prompt and/or self._functions."
-                    )
+                        f"Unknown function.")
                 fn = getattr(self, frame.function_name)
                 result = fn(json.loads(frame.arguments))
 
@@ -306,9 +306,9 @@ async def main(room_url: str, token):
 
         messages = []
 
-        llm = OpenAILLMService(
-            api_key=os.getenv("OPENAI_API_KEY"),
-            model="gpt-4-1106-preview",
+        llm = FireworksLLMService(
+            api_key=os.getenv("FIREWORKS_API_KEY"),
+            model="accounts/fireworks/models/firefunction-v1"
         )
         # tts = DeepgramTTSService(
         #     aiohttp_session=session,
@@ -330,7 +330,7 @@ async def main(room_url: str, token):
         pipeline = Pipeline(processors=[fl, llm, fl2, checklist, tts])
 
         @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        async def on_first_other_participant_joined(transport, participant):
             await pipeline.queue_frames([OpenAILLMContextFrame(context)])
 
         async def handle_intake():
@@ -340,8 +340,6 @@ async def main(room_url: str, token):
                 pre_processor=OpenAIUserContextAggregator(context),
             )
 
-        transport.transcription_settings["extra"]["endpointing"] = True
-        transport.transcription_settings["extra"]["punctuate"] = True
         try:
             await asyncio.gather(transport.run(), handle_intake())
         except (asyncio.CancelledError, KeyboardInterrupt):

examples/starter-apps/storybot.py

@@ -9,20 +9,21 @@ import wave
 from typing import AsyncGenerator
 from PIL import Image
 
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureLLMService, AzureTTSService
-from dailyai.services.fal_ai_services import FalImageGenService
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.services.deepgram_ai_services import DeepgramTTSService
-from dailyai.services.elevenlabs_ai_service import ElevenLabsTTSService
-from dailyai.pipeline.aggregators import (
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.frame_processor import FrameProcessor
+from pipecat.services.live_stream import LiveStream
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.azure_ai_services import AzureLLMService, AzureTTSService
+from pipecat.services.fal_ai_services import FalImageGenService
+from pipecat.services.open_ai_services import OpenAILLMService
+from pipecat.services.deepgram_ai_services import DeepgramTTSService
+from pipecat.services.elevenlabs_ai_services import ElevenLabsTTSService
+from pipecat.pipeline.aggregators import (
     LLMAssistantContextAggregator,
-    UserResponseAggregator,
-    LLMResponseAggregator,
+    LLMAssistantResponseAggregator,
+    LLMUserResponseAggregator,
 )
-from dailyai.pipeline.frames import (
+from pipecat.pipeline.frames import (
     EndPipeFrame,
     LLMMessagesFrame,
     Frame,
@@ -32,7 +33,7 @@ from dailyai.pipeline.frames import (
     ImageFrame,
     UserStoppedSpeakingFrame,
 )
-from dailyai.services.ai_services import FrameLogger, AIService
+from pipecat.services.ai_services import FrameLogger, AIService
 
 from runner import configure
 
@@ -40,7 +41,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 sounds = {}
@@ -99,7 +100,7 @@ class StoryProcessor(FrameProcessor):
         1. Catch the frames that are generated by the LLM service
         """
         if isinstance(frame, UserStoppedSpeakingFrame):
-            yield ImageFrame(None, images["grandma-writing.png"])
+            yield ImageFrame(images["grandma-writing.png"], (1024, 1024))
             yield AudioFrame(sounds["talking.wav"])
 
         elif isinstance(frame, TextFrame):
@@ -112,7 +113,7 @@ class StoryProcessor(FrameProcessor):
 
                 self._text = self._text.replace("\n", " ")
                 if len(self._text) > 2:
-                    yield ImageFrame(None, images["grandma-writing.png"])
+                    yield ImageFrame(images["grandma-writing.png"], (1024, 1024))
                     yield StoryStartFrame(self._text)
                     yield AudioFrame(sounds["ding3.wav"])
                 self._text = ""
@@ -146,11 +147,11 @@ class StoryProcessor(FrameProcessor):
                 # last bit
                 pass
         elif isinstance(frame, LLMResponseEndFrame):
-            yield ImageFrame(None, images["grandma-writing.png"])
+            yield ImageFrame(images["grandma-writing.png"], (1024, 1024))
             yield StoryPromptFrame(self._text)
             self._text = ""
             yield frame
-            yield ImageFrame(None, images["grandma-listening.png"])
+            yield ImageFrame(images["grandma-listening.png"], (1024, 1024))
             yield AudioFrame(sounds["listening.wav"])
 
         else:
@@ -204,13 +205,14 @@ async def main(room_url: str, token):
             voice_id="Xb7hH8MSUJpSbSDYk0k2",
         )  # matilda
         img = FalImageGenService(
-            image_size="1024x1024",
+            params={
+                image_size = "1024x1024",
+            },
             aiohttp_session=session,
-            key_id=os.getenv("FAL_KEY_ID"),
-            key_secret=os.getenv("FAL_KEY_SECRET"),
+            key=os.getenv("FAL_KEY"),
         )
-        lra = LLMResponseAggregator(messages)
-        ura = UserResponseAggregator(messages)
+        lra = LLMAssistantResponseAggregator(messages)
+        ura = LLMUserResponseAggregator(messages)
         sp = StoryProcessor(messages, story)
         sig = StoryImageGenerator(story, llm, img)
 
@@ -232,7 +234,7 @@ async def main(room_url: str, token):
         start_story_event = asyncio.Event()
 
         @transport.event_handler("on_first_other_participant_joined")
-        async def on_first_other_participant_joined(transport):
+        async def on_first_other_participant_joined(transport, participant):
             start_story_event.set()
 
         async def storytime():
@@ -252,7 +254,7 @@ async def main(room_url: str, token):
                 [llm, lca, tts], sink=transport.send_queue)
             await local_pipeline.queue_frames(
                 [
-                    ImageFrame(None, images["grandma-listening.png"]),
+                    ImageFrame(images["grandma-listening.png"], (1024, 1024)),
                     LLMMessagesFrame(intro_messages),
                     AudioFrame(sounds["listening.wav"]),
                     EndPipeFrame(),
@@ -260,6 +262,10 @@ async def main(room_url: str, token):
             )
             await local_pipeline.run_pipeline()
 
+            pipeline = Pipeline([llm, lca, tts, ls_sink])
+            pipeline.queue_frames([...])
+            pipeline.run()
+
             fl = FrameLogger("### After Image Generation")
             pipeline = Pipeline(
                 processors=[
@@ -276,8 +282,6 @@ async def main(room_url: str, token):
                 pipeline,
             )
 
-        transport.transcription_settings["extra"]["endpointing"] = True
-        transport.transcription_settings["extra"]["punctuate"] = True
         try:
             await asyncio.gather(transport.run(), storytime())
         except (asyncio.CancelledError, KeyboardInterrupt):

examples/starter-apps/translator.py

@@ -4,21 +4,21 @@ import logging
 import os
 from typing import AsyncGenerator
 
-from dailyai.pipeline.aggregators import (
+from pipecat.pipeline.aggregators import (
     SentenceAggregator,
 )
-from dailyai.pipeline.frames import (
+from pipecat.pipeline.frames import (
     Frame,
     LLMMessagesFrame,
     TextFrame,
     SendAppMessageFrame,
 )
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.transports.daily_transport import DailyTransport
-from dailyai.services.azure_ai_services import AzureTTSService
-from dailyai.services.open_ai_services import OpenAILLMService
-from dailyai.pipeline.aggregators import LLMFullResponseAggregator
+from pipecat.pipeline.frame_processor import FrameProcessor
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.transports.daily_transport import DailyTransport
+from pipecat.services.azure_ai_services import AzureTTSService
+from pipecat.services.open_ai_services import OpenAILLMService
+from pipecat.pipeline.aggregators import LLMFullResponseAggregator
 
 from runner import configure
 
@@ -28,7 +28,7 @@ from dotenv import load_dotenv
 load_dotenv(override=True)
 
 logging.basicConfig(format=f"%(levelno)s %(asctime)s %(message)s")
-logger = logging.getLogger("dailyai")
+logger = logging.getLogger("pipecat")
 logger.setLevel(logging.DEBUG)
 
 """
@@ -99,8 +99,6 @@ async def main(room_url: str, token):
         ts = TranslationSubtitles("spanish")
         pipeline = Pipeline([sa, tp, llm, lfra, ts, tts])
 
-        transport.transcription_settings["extra"]["endpointing"] = True
-        transport.transcription_settings["extra"]["punctuate"] = True
         await transport.run(pipeline)
 
 

linux-py3.10-requirements.txt

@@ -4,29 +4,28 @@
 #
 #    pip-compile --all-extras pyproject.toml
 #
-aiohttp==3.9.3
-    # via dailyai (pyproject.toml)
+aiohttp==3.9.5
+    # via pipecat (pyproject.toml)
 aiosignal==1.3.1
     # via aiohttp
-anthropic==0.20.0
-    # via dailyai (pyproject.toml)
+annotated-types==0.6.0
+    # via pydantic
+anthropic==0.25.8
+    # via pipecat (pyproject.toml)
 anyio==4.3.0
     # via
     #   anthropic
     #   httpx
     #   openai
-    #   starlette
 async-timeout==4.0.3
     # via aiohttp
 attrs==23.2.0
-    # via
-    #   aiohttp
-    #   fal
-av==11.0.0
+    # via aiohttp
+av==12.0.0
     # via faster-whisper
-azure-cognitiveservices-speech==1.36.0
-    # via dailyai (pyproject.toml)
-blinker==1.7.0
+azure-cognitiveservices-speech==1.37.0
+    # via pipecat (pyproject.toml)
+blinker==1.8.2
     # via flask
 certifi==2024.2.2
     # via
@@ -36,48 +35,38 @@ certifi==2024.2.2
 charset-normalizer==3.3.2
     # via requests
 click==8.1.7
-    # via
-    #   fal
-    #   flask
-colorama==0.4.6
-    # via fal
+    # via flask
 coloredlogs==15.0.1
     # via onnxruntime
-ctranslate2==4.1.0
+ctranslate2==4.2.1
     # via faster-whisper
-daily-python==0.7.2
-    # via dailyai (pyproject.toml)
-deprecated==1.2.14
-    # via opentelemetry-api
-dill==0.3.7
-    # via fal
-distlib==0.3.8
-    # via virtualenv
+daily-python==0.7.4
+    # via pipecat (pyproject.toml)
 distro==1.9.0
     # via
     #   anthropic
     #   openai
-exceptiongroup==1.2.0
+einops==0.8.0
+    # via pipecat (pyproject.toml)
+exceptiongroup==1.2.1
     # via anyio
-fal==0.12.3
-    # via dailyai (pyproject.toml)
-fastapi==0.99.1
-    # via fal
-faster-whisper==1.0.1
-    # via dailyai (pyproject.toml)
-filelock==3.13.3
+fal-client==0.4.0
+    # via pipecat (pyproject.toml)
+faster-whisper==1.0.2
+    # via pipecat (pyproject.toml)
+filelock==3.14.0
     # via
     #   huggingface-hub
     #   pyht
     #   torch
+    #   transformers
     #   triton
-    #   virtualenv
-flask==3.0.2
+flask==3.0.3
     # via
-    #   dailyai (pyproject.toml)
     #   flask-cors
-flask-cors==4.0.0
-    # via dailyai (pyproject.toml)
+    #   pipecat (pyproject.toml)
+flask-cors==4.0.1
+    # via pipecat (pyproject.toml)
 flatbuffers==24.3.25
     # via onnxruntime
 frozenlist==1.4.1
@@ -88,15 +77,8 @@ fsspec==2024.3.1
     # via
     #   huggingface-hub
     #   torch
-grpc-interceptor==0.15.4
-    # via fal
-grpcio==1.62.1
-    # via
-    #   fal
-    #   grpc-interceptor
-    #   isolate
-    #   isolate-proto
-    #   pyht
+grpcio==1.63.0
+    # via pyht
 h11==0.14.0
     # via httpcore
 httpcore==1.0.5
@@ -104,57 +86,51 @@ httpcore==1.0.5
 httpx==0.27.0
     # via
     #   anthropic
-    #   fal
+    #   fal-client
     #   openai
-huggingface-hub==0.22.2
+httpx-sse==0.4.0
+    # via fal-client
+huggingface-hub==0.23.0
     # via
     #   faster-whisper
+    #   timm
     #   tokenizers
+    #   transformers
 humanfriendly==10.0
     # via coloredlogs
-idna==3.6
+idna==3.7
     # via
     #   anyio
     #   httpx
     #   requests
     #   yarl
-importlib-metadata==7.0.0
-    # via opentelemetry-api
-isolate[build]==0.12.7
-    # via
-    #   fal
-    #   isolate-proto
-isolate-proto==0.3.3
-    # via fal
-itsdangerous==2.1.2
+itsdangerous==2.2.0
     # via flask
-jinja2==3.1.3
+jinja2==3.1.4
     # via
     #   flask
     #   torch
-markdown-it-py==3.0.0
-    # via rich
+loguru==0.7.2
+    # via pipecat (pyproject.toml)
 markupsafe==2.1.5
     # via
     #   jinja2
     #   werkzeug
-mdurl==0.1.2
-    # via markdown-it-py
 mpmath==1.3.0
     # via sympy
-msgpack==1.0.8
-    # via fal
 multidict==6.0.5
     # via
     #   aiohttp
     #   yarl
-networkx==3.2.1
+networkx==3.3
     # via torch
 numpy==1.26.4
     # via
     #   ctranslate2
-    #   dailyai (pyproject.toml)
     #   onnxruntime
+    #   pipecat (pyproject.toml)
+    #   torchvision
+    #   transformers
 nvidia-cublas-cu12==12.1.3.1
     # via
     #   nvidia-cudnn-cu12
@@ -178,7 +154,7 @@ nvidia-cusparse-cu12==12.1.0.106
     # via
     #   nvidia-cusolver-cu12
     #   torch
-nvidia-nccl-cu12==2.19.3
+nvidia-nccl-cu12==2.20.5
     # via torch
 nvidia-nvjitlink-cu12==12.4.127
     # via
@@ -186,134 +162,106 @@ nvidia-nvjitlink-cu12==12.4.127
     #   nvidia-cusparse-cu12
 nvidia-nvtx-cu12==12.1.105
     # via torch
-onnxruntime==1.17.1
+onnxruntime==1.17.3
     # via faster-whisper
-openai==1.14.2
-    # via dailyai (pyproject.toml)
-opentelemetry-api==1.24.0
-    # via
-    #   fal
-    #   opentelemetry-sdk
-opentelemetry-sdk==1.24.0
-    # via fal
-opentelemetry-semantic-conventions==0.45b0
-    # via opentelemetry-sdk
+openai==1.26.0
+    # via pipecat (pyproject.toml)
 packaging==24.0
     # via
-    #   fal
     #   huggingface-hub
     #   onnxruntime
-pathspec==0.11.2
-    # via fal
-pillow==10.2.0
+    #   transformers
+pillow==10.3.0
     # via
-    #   dailyai (pyproject.toml)
-    #   fal
-platformdirs==4.2.0
-    # via
-    #   isolate
-    #   virtualenv
-portalocker==2.8.2
-    # via fal
+    #   pipecat (pyproject.toml)
+    #   torchvision
 protobuf==4.25.3
     # via
-    #   isolate
-    #   isolate-proto
     #   onnxruntime
     #   pyht
 pyaudio==0.2.14
-    # via dailyai (pyproject.toml)
-pydantic==1.10.15
+    # via pipecat (pyproject.toml)
+pydantic==2.7.1
     # via
     #   anthropic
-    #   fal
-    #   fastapi
     #   openai
-pygments==2.17.2
-    # via rich
-pyht==0.0.26
-    # via dailyai (pyproject.toml)
-pyjwt==2.8.0
-    # via fal
-python-dateutil==2.9.0.post0
-    # via fal
+pydantic-core==2.18.2
+    # via pydantic
+pyht==0.0.28
+    # via pipecat (pyproject.toml)
 python-dotenv==1.0.1
-    # via dailyai (pyproject.toml)
+    # via pipecat (pyproject.toml)
 pyyaml==6.0.1
     # via
     #   ctranslate2
     #   huggingface-hub
-    #   isolate
+    #   timm
+    #   transformers
+regex==2024.5.10
+    # via transformers
 requests==2.31.0
     # via
     #   huggingface-hub
     #   pyht
-rich==13.7.1
-    # via fal
-six==1.16.0
-    # via python-dateutil
+    #   transformers
+safetensors==0.4.3
+    # via
+    #   timm
+    #   transformers
 sniffio==1.3.1
     # via
     #   anthropic
     #   anyio
     #   httpx
     #   openai
-starlette==0.27.0
-    # via fastapi
-structlog==22.3.0
-    # via fal
 sympy==1.12
     # via
     #   onnxruntime
     #   torch
-tblib==3.0.0
-    # via isolate
-tokenizers==0.15.2
+timm==0.9.16
+    # via pipecat (pyproject.toml)
+tokenizers==0.19.1
     # via
     #   anthropic
     #   faster-whisper
-torch==2.2.1
+    #   transformers
+torch==2.3.0
     # via
-    #   dailyai (pyproject.toml)
+    #   pipecat (pyproject.toml)
+    #   timm
     #   torchaudio
-torchaudio==2.2.1
-    # via dailyai (pyproject.toml)
-tqdm==4.66.2
+    #   torchvision
+torchaudio==2.3.0
+    # via pipecat (pyproject.toml)
+torchvision==0.18.0
+    # via timm
+tqdm==4.66.4
     # via
     #   huggingface-hub
     #   openai
-triton==2.2.0
+    #   transformers
+transformers==4.40.2
+    # via pipecat (pyproject.toml)
+triton==2.3.0
     # via torch
-types-python-dateutil==2.9.0.20240316
-    # via fal
-typing-extensions==4.10.0
+typing-extensions==4.11.0
     # via
     #   anthropic
     #   anyio
-    #   dailyai (pyproject.toml)
-    #   fal
-    #   fastapi
     #   huggingface-hub
     #   openai
-    #   opentelemetry-sdk
+    #   pipecat (pyproject.toml)
     #   pydantic
+    #   pydantic-core
     #   torch
 urllib3==2.2.1
     # via requests
-virtualenv==20.25.1
-    # via isolate
 websockets==12.0
-    # via
-    #   dailyai (pyproject.toml)
-    #   fal
-werkzeug==3.0.2
+    # via pipecat (pyproject.toml)
+werkzeug==3.0.3
     # via flask
-wrapt==1.16.0
-    # via deprecated
 yarl==1.9.4
     # via aiohttp
-zipp==3.18.1
-    # via importlib-metadata
 
 # The following packages are considered to be unsafe in a requirements file:
 # setuptools

macos-py3.10-requirements.txt

@@ -4,29 +4,28 @@
 #
 #    pip-compile --all-extras pyproject.toml
 #
-aiohttp==3.9.3
-    # via dailyai (pyproject.toml)
+aiohttp==3.9.5
+    # via pipecat (pyproject.toml)
 aiosignal==1.3.1
     # via aiohttp
-anthropic==0.20.0
-    # via dailyai (pyproject.toml)
+annotated-types==0.6.0
+    # via pydantic
+anthropic==0.25.8
+    # via pipecat (pyproject.toml)
 anyio==4.3.0
     # via
     #   anthropic
     #   httpx
     #   openai
-    #   starlette
 async-timeout==4.0.3
     # via aiohttp
 attrs==23.2.0
-    # via
-    #   aiohttp
-    #   fal
-av==11.0.0
+    # via aiohttp
+av==12.0.0
     # via faster-whisper
-azure-cognitiveservices-speech==1.36.0
-    # via dailyai (pyproject.toml)
-blinker==1.7.0
+azure-cognitiveservices-speech==1.37.0
+    # via pipecat (pyproject.toml)
+blinker==1.8.2
     # via flask
 certifi==2024.2.2
     # via
@@ -36,47 +35,37 @@ certifi==2024.2.2
 charset-normalizer==3.3.2
     # via requests
 click==8.1.7
-    # via
-    #   fal
-    #   flask
-colorama==0.4.6
-    # via fal
+    # via flask
 coloredlogs==15.0.1
     # via onnxruntime
-ctranslate2==4.1.0
+ctranslate2==4.2.1
     # via faster-whisper
-daily-python==0.7.2
-    # via dailyai (pyproject.toml)
-deprecated==1.2.14
-    # via opentelemetry-api
-dill==0.3.7
-    # via fal
-distlib==0.3.8
-    # via virtualenv
+daily-python==0.7.4
+    # via pipecat (pyproject.toml)
 distro==1.9.0
     # via
     #   anthropic
     #   openai
-exceptiongroup==1.2.0
+einops==0.8.0
+    # via pipecat (pyproject.toml)
+exceptiongroup==1.2.1
     # via anyio
-fal==0.12.3
-    # via dailyai (pyproject.toml)
-fastapi==0.99.1
-    # via fal
-faster-whisper==1.0.1
-    # via dailyai (pyproject.toml)
-filelock==3.13.3
+fal-client==0.4.0
+    # via pipecat (pyproject.toml)
+faster-whisper==1.0.2
+    # via pipecat (pyproject.toml)
+filelock==3.14.0
     # via
     #   huggingface-hub
     #   pyht
     #   torch
-    #   virtualenv
-flask==3.0.2
+    #   transformers
+flask==3.0.3
     # via
-    #   dailyai (pyproject.toml)
     #   flask-cors
-flask-cors==4.0.0
-    # via dailyai (pyproject.toml)
+    #   pipecat (pyproject.toml)
+flask-cors==4.0.1
+    # via pipecat (pyproject.toml)
 flatbuffers==24.3.25
     # via onnxruntime
 frozenlist==1.4.1
@@ -87,15 +76,8 @@ fsspec==2024.3.1
     # via
     #   huggingface-hub
     #   torch
-grpc-interceptor==0.15.4
-    # via fal
-grpcio==1.62.1
-    # via
-    #   fal
-    #   grpc-interceptor
-    #   isolate
-    #   isolate-proto
-    #   pyht
+grpcio==1.63.0
+    # via pyht
 h11==0.14.0
     # via httpcore
 httpcore==1.0.5
@@ -103,183 +85,149 @@ httpcore==1.0.5
 httpx==0.27.0
     # via
     #   anthropic
-    #   fal
+    #   fal-client
     #   openai
-huggingface-hub==0.22.2
+httpx-sse==0.4.0
+    # via fal-client
+huggingface-hub==0.23.0
     # via
     #   faster-whisper
+    #   timm
     #   tokenizers
+    #   transformers
 humanfriendly==10.0
     # via coloredlogs
-idna==3.6
+idna==3.7
     # via
     #   anyio
     #   httpx
     #   requests
     #   yarl
-importlib-metadata==7.0.0
-    # via opentelemetry-api
-isolate[build]==0.12.7
-    # via
-    #   fal
-    #   isolate-proto
-isolate-proto==0.3.3
-    # via fal
-itsdangerous==2.1.2
+itsdangerous==2.2.0
     # via flask
-jinja2==3.1.3
+jinja2==3.1.4
     # via
     #   flask
     #   torch
-markdown-it-py==3.0.0
-    # via rich
+loguru==0.7.2
+    # via pipecat (pyproject.toml)
 markupsafe==2.1.5
     # via
     #   jinja2
     #   werkzeug
-mdurl==0.1.2
-    # via markdown-it-py
 mpmath==1.3.0
     # via sympy
-msgpack==1.0.8
-    # via fal
 multidict==6.0.5
     # via
     #   aiohttp
     #   yarl
-networkx==3.2.1
+networkx==3.3
     # via torch
 numpy==1.26.4
     # via
     #   ctranslate2
-    #   dailyai (pyproject.toml)
     #   onnxruntime
-onnxruntime==1.17.1
+    #   pipecat (pyproject.toml)
+    #   torchvision
+    #   transformers
+onnxruntime==1.17.3
     # via faster-whisper
-openai==1.14.2
-    # via dailyai (pyproject.toml)
-opentelemetry-api==1.24.0
-    # via
-    #   fal
-    #   opentelemetry-sdk
-opentelemetry-sdk==1.24.0
-    # via fal
-opentelemetry-semantic-conventions==0.45b0
-    # via opentelemetry-sdk
+openai==1.26.0
+    # via pipecat (pyproject.toml)
 packaging==24.0
     # via
-    #   fal
     #   huggingface-hub
     #   onnxruntime
-pathspec==0.11.2
-    # via fal
-pillow==10.2.0
+    #   transformers
+pillow==10.3.0
     # via
-    #   dailyai (pyproject.toml)
-    #   fal
-platformdirs==4.2.0
-    # via
-    #   isolate
-    #   virtualenv
-portalocker==2.8.2
-    # via fal
+    #   pipecat (pyproject.toml)
+    #   torchvision
 protobuf==4.25.3
     # via
-    #   isolate
-    #   isolate-proto
     #   onnxruntime
     #   pyht
 pyaudio==0.2.14
-    # via dailyai (pyproject.toml)
-pydantic==1.10.15
+    # via pipecat (pyproject.toml)
+pydantic==2.7.1
     # via
     #   anthropic
-    #   fal
-    #   fastapi
     #   openai
-pygments==2.17.2
-    # via rich
-pyht==0.0.26
-    # via dailyai (pyproject.toml)
-pyjwt==2.8.0
-    # via fal
-python-dateutil==2.9.0.post0
-    # via fal
+pydantic-core==2.18.2
+    # via pydantic
+pyht==0.0.28
+    # via pipecat (pyproject.toml)
 python-dotenv==1.0.1
-    # via dailyai (pyproject.toml)
+    # via pipecat (pyproject.toml)
 pyyaml==6.0.1
     # via
     #   ctranslate2
     #   huggingface-hub
-    #   isolate
+    #   timm
+    #   transformers
+regex==2024.5.10
+    # via transformers
 requests==2.31.0
     # via
     #   huggingface-hub
     #   pyht
-rich==13.7.1
-    # via fal
-six==1.16.0
-    # via python-dateutil
+    #   transformers
+safetensors==0.4.3
+    # via
+    #   timm
+    #   transformers
 sniffio==1.3.1
     # via
     #   anthropic
     #   anyio
     #   httpx
     #   openai
-starlette==0.27.0
-    # via fastapi
-structlog==22.3.0
-    # via fal
 sympy==1.12
     # via
     #   onnxruntime
     #   torch
-tblib==3.0.0
-    # via isolate
-tokenizers==0.15.2
+timm==0.9.16
+    # via pipecat (pyproject.toml)
+tokenizers==0.19.1
     # via
     #   anthropic
     #   faster-whisper
-torch==2.2.1
+    #   transformers
+torch==2.3.0
     # via
-    #   dailyai (pyproject.toml)
+    #   pipecat (pyproject.toml)
+    #   timm
     #   torchaudio
-torchaudio==2.2.1
-    # via dailyai (pyproject.toml)
-tqdm==4.66.2
+    #   torchvision
+torchaudio==2.3.0
+    # via pipecat (pyproject.toml)
+torchvision==0.18.0
+    # via timm
+tqdm==4.66.4
     # via
     #   huggingface-hub
     #   openai
-types-python-dateutil==2.9.0.20240316
-    # via fal
-typing-extensions==4.10.0
+    #   transformers
+transformers==4.40.2
+    # via pipecat (pyproject.toml)
+typing-extensions==4.11.0
     # via
     #   anthropic
     #   anyio
-    #   dailyai (pyproject.toml)
-    #   fal
-    #   fastapi
     #   huggingface-hub
     #   openai
-    #   opentelemetry-sdk
+    #   pipecat (pyproject.toml)
     #   pydantic
+    #   pydantic-core
     #   torch
 urllib3==2.2.1
     # via requests
-virtualenv==20.25.1
-    # via isolate
 websockets==12.0
-    # via
-    #   dailyai (pyproject.toml)
-    #   fal
-werkzeug==3.0.2
+    # via pipecat (pyproject.toml)
+werkzeug==3.0.3
     # via flask
-wrapt==1.16.0
-    # via deprecated
 yarl==1.9.4
     # via aiohttp
-zipp==3.18.1
-    # via importlib-metadata
 
 # The following packages are considered to be unsafe in a requirements file:
 # setuptools

pyproject.toml

@@ -3,9 +3,9 @@ requires = ["setuptools>=64", "setuptools_scm>=8"]
 build-backend = "setuptools.build_meta"
 
 [project]
-name = "dailyai"
+name = "pipecat-ai"
 dynamic = ["version"]
-description = "An open source framework for real-time, multi-modal, conversational AI applications"
+description = "An open source framework for voice (and multimodal) assistants"
 license = { text = "BSD 2-Clause License" }
 readme = "README.md"
 requires-python = ">=3.7"
@@ -20,28 +20,31 @@ classifiers = [
     "Topic :: Scientific/Engineering :: Artificial Intelligence"
 ]
 dependencies = [
-    "aiohttp==3.9.3",
-    "numpy==1.26.4",
-    "Pillow==10.2.0",
-    "typing-extensions==4.10.0",
+    "aiohttp~=3.9.5",
+    "numpy~=1.26.4",
+    "loguru~=0.7.0",
+    "Pillow~=10.3.0",
+    "typing-extensions~=4.11.0",
 ]
 
 [project.urls]
-Source = "https://github.com/daily-co/dailyai"
-Website = "https://daily.co"
+Source = "https://github.com/pipecat-ai/pipecat"
+Website = "https://pipecat.ai"
 
 [project.optional-dependencies]
-anthropic = [ "anthropic==0.20.0" ]
-azure = [ "azure-cognitiveservices-speech==1.36.0" ]
-daily = [ "daily-python==0.7.2" ]
-examples = [ "python-dotenv==1.0.1", "flask==3.0.2", "flask_cors==4.0.0" ]
-fal = [ "fal==0.12.3" ]
-local = [ "pyaudio==0.2.14" ]
-openai = [ "openai==1.14.2" ]
-playht = [ "pyht==0.0.26" ]
-silero = [ "torch==2.2.1", "torchaudio==2.2.1" ]
-websocket = [ "websockets==12.0" ]
-whisper = [ "faster_whisper==1.0.1" ]
+anthropic = [ "anthropic~=0.25.7" ]
+azure = [ "azure-cognitiveservices-speech~=1.37.0" ]
+daily = [ "daily-python~=0.7.4" ]
+examples = [ "python-dotenv~=1.0.0", "flask~=3.0.3", "flask_cors~=4.0.1" ]
+fal = [ "fal-client~=0.4.0" ]
+fireworks = [ "openai~=1.26.0" ]
+local = [ "pyaudio~=0.2.0" ]
+moondream = [ "einops~=0.8.0", "timm~=0.9.16", "transformers~=4.40.2" ]
+openai = [ "openai~=1.26.0" ]
+playht = [ "pyht~=0.0.28" ]
+silero = [ "torch~=2.3.0", "torchaudio~=2.3.0" ]
+websocket = [ "websockets~=12.0" ]
+whisper = [ "faster-whisper~=1.0.2" ]
 
 [tool.setuptools.packages.find]
 # All the following settings are optional:
@@ -51,4 +54,4 @@ where = ["src"]
 pythonpath = ["src"]
 
 [tool.setuptools_scm]
-# Empty
+local_scheme = "no-local-version"

src/dailyai/pipeline/aggregators.py

@@ -1,392 +0,0 @@
-import asyncio
-import re
-
-from dailyai.pipeline.frame_processor import FrameProcessor
-
-from dailyai.pipeline.frames import (
-    EndFrame,
-    EndPipeFrame,
-    Frame,
-    LLMMessagesFrame,
-    LLMResponseEndFrame,
-    LLMResponseStartFrame,
-    TextFrame,
-    TranscriptionFrame,
-    UserStartedSpeakingFrame,
-    UserStoppedSpeakingFrame,
-)
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.services.ai_services import AIService
-
-from typing import AsyncGenerator, Coroutine, List
-
-
-class ResponseAggregator(FrameProcessor):
-
-    def __init__(
-        self,
-        *,
-        messages: list[dict] | None,
-        role: str,
-        start_frame,
-        end_frame,
-        accumulator_frame,
-        pass_through=True,
-    ):
-        self.aggregation = ""
-        self.aggregating = False
-        self.messages = messages
-        self._role = role
-        self._start_frame = start_frame
-        self._end_frame = end_frame
-        self._accumulator_frame = accumulator_frame
-        self._pass_through = pass_through
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if not self.messages:
-            return
-
-        if isinstance(frame, self._start_frame):
-            self.aggregating = True
-        elif isinstance(frame, self._end_frame):
-            self.aggregating = False
-            # Sometimes VAD triggers quickly on and off. If we don't get any transcription,
-            # it creates empty LLM message queue frames
-            if len(self.aggregation) > 0:
-                self.messages.append(
-                    {"role": self._role, "content": self.aggregation})
-                self.aggregation = ""
-                yield self._end_frame()
-                yield LLMMessagesFrame(self.messages)
-        elif isinstance(frame, self._accumulator_frame) and self.aggregating:
-            self.aggregation += f" {frame.text}"
-            if self._pass_through:
-                yield frame
-        else:
-            yield frame
-
-
-class LLMResponseAggregator(ResponseAggregator):
-    def __init__(self, messages: list[dict]):
-        super().__init__(
-            messages=messages,
-            role="assistant",
-            start_frame=LLMResponseStartFrame,
-            end_frame=LLMResponseEndFrame,
-            accumulator_frame=TextFrame,
-        )
-
-
-class UserResponseAggregator(ResponseAggregator):
-    def __init__(self, messages: list[dict]):
-        super().__init__(
-            messages=messages,
-            role="user",
-            start_frame=UserStartedSpeakingFrame,
-            end_frame=UserStoppedSpeakingFrame,
-            accumulator_frame=TranscriptionFrame,
-            pass_through=False,
-        )
-
-
-class LLMContextAggregator(AIService):
-    def __init__(
-        self,
-        messages: list[dict],
-        role: str,
-        bot_participant_id=None,
-        complete_sentences=True,
-        pass_through=True,
-    ):
-        super().__init__()
-        self.messages = messages
-        self.bot_participant_id = bot_participant_id
-        self.role = role
-        self.sentence = ""
-        self.complete_sentences = complete_sentences
-        self.pass_through = pass_through
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        # We don't do anything with non-text frames, pass it along to next in
-        # the pipeline.
-        if not isinstance(frame, TextFrame):
-            yield frame
-            return
-
-        # Ignore transcription frames from the bot
-        if isinstance(frame, TranscriptionFrame):
-            if frame.participantId == self.bot_participant_id:
-                return
-
-        # The common case for "pass through" is receiving frames from the LLM that we'll
-        # use to update the "assistant" LLM messages, but also passing the text frames
-        # along to a TTS service to be spoken to the user.
-        if self.pass_through:
-            yield frame
-
-        # TODO: split up transcription by participant
-        if self.complete_sentences:
-            # type: ignore -- the linter thinks this isn't a TextFrame, even
-            # though we check it above
-            self.sentence += frame.text
-            if self.sentence.endswith((".", "?", "!")):
-                self.messages.append(
-                    {"role": self.role, "content": self.sentence})
-                self.sentence = ""
-                yield LLMMessagesFrame(self.messages)
-        else:
-            # type: ignore -- the linter thinks this isn't a TextFrame, even
-            # though we check it above
-            self.messages.append({"role": self.role, "content": frame.text})
-            yield LLMMessagesFrame(self.messages)
-
-
-class LLMUserContextAggregator(LLMContextAggregator):
-    def __init__(
-            self,
-            messages: list[dict],
-            bot_participant_id=None,
-            complete_sentences=True):
-        super().__init__(
-            messages,
-            "user",
-            bot_participant_id,
-            complete_sentences,
-            pass_through=False)
-
-
-class LLMAssistantContextAggregator(LLMContextAggregator):
-    def __init__(
-            self,
-            messages: list[dict],
-            bot_participant_id=None,
-            complete_sentences=True):
-        super().__init__(
-            messages,
-            "assistant",
-            bot_participant_id,
-            complete_sentences,
-            pass_through=True,
-        )
-
-
-class SentenceAggregator(FrameProcessor):
-    """This frame processor aggregates text frames into complete sentences.
-
-    Frame input/output:
-        TextFrame("Hello,") -> None
-        TextFrame(" world.") -> TextFrame("Hello world.")
-
-    Doctest:
-    >>> async def print_frames(aggregator, frame):
-    ...     async for frame in aggregator.process_frame(frame):
-    ...         print(frame.text)
-
-    >>> aggregator = SentenceAggregator()
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello,")))
-    >>> asyncio.run(print_frames(aggregator, TextFrame(" world.")))
-    Hello, world.
-    """
-
-    def __init__(self):
-        self.aggregation = ""
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, TextFrame):
-            m = re.search("(.*[?.!])(.*)", frame.text)
-            if m:
-                yield TextFrame(self.aggregation + m.group(1))
-                self.aggregation = m.group(2)
-            else:
-                self.aggregation += frame.text
-        elif isinstance(frame, EndFrame):
-            if self.aggregation:
-                yield TextFrame(self.aggregation)
-            yield frame
-        else:
-            yield frame
-
-
-class LLMFullResponseAggregator(FrameProcessor):
-    """This class aggregates Text frames until it receives a
-    LLMResponseEndFrame, then emits the concatenated text as
-    a single text frame.
-
-    given the following frames:
-
-        TextFrame("Hello,")
-        TextFrame(" world.")
-        TextFrame(" I am")
-        TextFrame(" an LLM.")
-        LLMResponseEndFrame()]
-
-    this processor will yield nothing for the first 4 frames, then
-
-        TextFrame("Hello, world. I am an LLM.")
-        LLMResponseEndFrame()
-
-    when passed the last frame.
-
-    >>> async def print_frames(aggregator, frame):
-    ...     async for frame in aggregator.process_frame(frame):
-    ...         if isinstance(frame, TextFrame):
-    ...             print(frame.text)
-    ...         else:
-    ...             print(frame.__class__.__name__)
-
-    >>> aggregator = LLMFullResponseAggregator()
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello,")))
-    >>> asyncio.run(print_frames(aggregator, TextFrame(" world.")))
-    >>> asyncio.run(print_frames(aggregator, TextFrame(" I am")))
-    >>> asyncio.run(print_frames(aggregator, TextFrame(" an LLM.")))
-    >>> asyncio.run(print_frames(aggregator, LLMResponseEndFrame()))
-    Hello, world. I am an LLM.
-    LLMResponseEndFrame
-    """
-
-    def __init__(self):
-        self.aggregation = ""
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, TextFrame):
-            self.aggregation += frame.text
-        elif isinstance(frame, LLMResponseEndFrame):
-            yield TextFrame(self.aggregation)
-            yield frame
-            self.aggregation = ""
-        else:
-            yield frame
-
-
-class StatelessTextTransformer(FrameProcessor):
-    """This processor calls the given function on any text in a text frame.
-
-    >>> async def print_frames(aggregator, frame):
-    ...     async for frame in aggregator.process_frame(frame):
-    ...         print(frame.text)
-
-    >>> aggregator = StatelessTextTransformer(lambda x: x.upper())
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello")))
-    HELLO
-    """
-
-    def __init__(self, transform_fn):
-        self.transform_fn = transform_fn
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, TextFrame):
-            result = self.transform_fn(frame.text)
-            if isinstance(result, Coroutine):
-                result = await result
-
-            yield TextFrame(result)
-        else:
-            yield frame
-
-
-class ParallelPipeline(FrameProcessor):
-    """Run multiple pipelines in parallel.
-
-    This class takes frames from its source queue and sends them to each
-    sub-pipeline. Each sub-pipeline emits its frames into this class's
-    sink queue. No guarantees are made about the ordering of frames in
-    the sink queue (that is, no sub-pipeline has higher priority than
-    any other, frames are put on the sink in the order they're emitted
-    by the sub-pipelines).
-
-    After each frame is taken from this class's source queue and placed
-    in each sub-pipeline's source queue, an EndPipeFrame is put on each
-    sub-pipeline's source queue. This indicates to the sub-pipe runner
-    that it should exit.
-
-    Since frame handlers pass through unhandled frames by convention, this
-    class de-dupes frames in its sink before yielding them.
-    """
-
-    def __init__(self, pipeline_definitions: List[List[FrameProcessor]]):
-        self.sources = [asyncio.Queue() for _ in pipeline_definitions]
-        self.sink: asyncio.Queue[Frame] = asyncio.Queue()
-        self.pipelines: list[Pipeline] = [
-            Pipeline(
-                pipeline_definition,
-                source,
-                self.sink,
-            )
-            for source, pipeline_definition in zip(self.sources, pipeline_definitions)
-        ]
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        for source in self.sources:
-            await source.put(frame)
-            await source.put(EndPipeFrame())
-
-        await asyncio.gather(*[pipeline.run_pipeline() for pipeline in self.pipelines])
-
-        seen_ids = set()
-        while not self.sink.empty():
-            frame = await self.sink.get()
-
-            # de-dup frames. Because the convention is to yield a frame that isn't processed,
-            # each pipeline will likely yield the same frame, so we will end up with _n_ copies
-            # of unprocessed frames where _n_ is the number of parallel pipes that don't
-            # process that frame.
-            if id(frame) in seen_ids:
-                continue
-            seen_ids.add(id(frame))
-
-            # Skip passing along EndPipeFrame, because we use them
-            # for our own flow control.
-            if not isinstance(frame, EndPipeFrame):
-                yield frame
-
-
-class GatedAggregator(FrameProcessor):
-    """Accumulate frames, with custom functions to start and stop accumulation.
-    Yields gate-opening frame before any accumulated frames, then ensuing frames
-    until and not including the gate-closed frame.
-
-    >>> from dailyai.pipeline.frames import ImageFrame
-
-    >>> async def print_frames(aggregator, frame):
-    ...     async for frame in aggregator.process_frame(frame):
-    ...         if isinstance(frame, TextFrame):
-    ...             print(frame.text)
-    ...         else:
-    ...             print(frame.__class__.__name__)
-
-    >>> aggregator = GatedAggregator(
-    ...     gate_close_fn=lambda x: isinstance(x, LLMResponseStartFrame),
-    ...     gate_open_fn=lambda x: isinstance(x, ImageFrame),
-    ...     start_open=False)
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello")))
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello again.")))
-    >>> asyncio.run(print_frames(aggregator, ImageFrame(url='', image=bytes([]))))
-    ImageFrame
-    Hello
-    Hello again.
-    >>> asyncio.run(print_frames(aggregator, TextFrame("Goodbye.")))
-    Goodbye.
-    """
-
-    def __init__(self, gate_open_fn, gate_close_fn, start_open):
-        self.gate_open_fn = gate_open_fn
-        self.gate_close_fn = gate_close_fn
-        self.gate_open = start_open
-        self.accumulator: List[Frame] = []
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if self.gate_open:
-            if self.gate_close_fn(frame):
-                self.gate_open = False
-        else:
-            if self.gate_open_fn(frame):
-                self.gate_open = True
-
-        if self.gate_open:
-            yield frame
-            if self.accumulator:
-                for frame in self.accumulator:
-                    yield frame
-            self.accumulator = []
-        else:
-            self.accumulator.append(frame)

src/dailyai/pipeline/frame_processor.py

@@ -1,34 +0,0 @@
-from abc import abstractmethod
-from typing import AsyncGenerator
-
-from dailyai.pipeline.frames import ControlFrame, Frame
-
-
-class FrameProcessor:
-    """This is the base class for all frame processors. Frame processors consume a frame
-    and yield 0 or more frames. Generally frame processors are used as part of a pipeline
-    where frames come from a source queue, are processed by a series of frame processors,
-    then placed on a sink queue.
-
-    By convention, FrameProcessors should immediately yield any frames they don't process.
-
-    Stateful FrameProcessors should watch for the EndFrame and finalize their
-    output, eg. yielding an unfinished sentence if they're aggregating LLM output to full
-    sentences. EndFrame is also a chance to clean up any services that need to
-    be closed, del'd, etc.
-    """
-
-    @abstractmethod
-    async def process_frame(
-        self, frame: Frame
-    ) -> AsyncGenerator[Frame, None]:
-        """Process a single frame and yield 0 or more frames."""
-        yield frame
-
-    @abstractmethod
-    async def interrupted(self) -> None:
-        """Handle any cleanup if the pipeline was interrupted."""
-        pass
-
-    def __str__(self):
-        return self.__class__.__name__

src/dailyai/pipeline/frames.py

@@ -1,187 +0,0 @@
-from dataclasses import dataclass
-from typing import Any, List
-
-
-class Frame:
-    def __str__(self):
-        return f"{self.__class__.__name__}"
-
-
-class ControlFrame(Frame):
-    # Control frames should contain no instance data, so
-    # equality is based solely on the class.
-    def __eq__(self, other):
-        return isinstance(other, self.__class__)
-
-
-class StartFrame(ControlFrame):
-    """Used (but not required) to start a pipeline, and is also used to
-    indicate that an interruption has ended and the transport should start
-    processing frames again."""
-    pass
-
-
-class EndFrame(ControlFrame):
-    """Indicates that a pipeline has ended and frame processors and pipelines
-    should be shut down. If the transport receives this frame, it will stop
-    sending frames to its output channel(s) and close all its threads."""
-    pass
-
-
-class EndPipeFrame(ControlFrame):
-    """Indicates that a pipeline has ended but that the transport should
-    continue processing. This frame is used in parallel pipelines and other
-    sub-pipelines."""
-    pass
-
-
-class PipelineStartedFrame(ControlFrame):
-    """
-    Used by the transport to indicate that execution of a pipeline is starting
-    (or restarting). It should be the first frame your app receives when it
-    starts, or when an interruptible pipeline has been interrupted.
-    """
-
-    pass
-
-
-class LLMResponseStartFrame(ControlFrame):
-    """Used to indicate the beginning of an LLM response. Following TextFrames
-    are part of the LLM response until an LLMResponseEndFrame"""
-    pass
-
-
-class LLMResponseEndFrame(ControlFrame):
-    """Indicates the end of an LLM response."""
-    pass
-
-
-@dataclass()
-class AudioFrame(Frame):
-    """A chunk of audio. Will be played by the transport if the transport's mic
-    has been enabled."""
-    data: bytes
-
-    def __str__(self):
-        return f"{self.__class__.__name__}, size: {len(self.data)} B"
-
-
-@dataclass()
-class ImageFrame(Frame):
-    """An image. Will be shown by the transport if the transport's camera is
-    enabled."""
-    url: str | None
-    image: bytes
-
-    def __str__(self):
-        return f"{self.__class__.__name__}, url: {self.url}, image size: {len(self.image)} B"
-
-
-@dataclass()
-class SpriteFrame(Frame):
-    """An animated sprite. Will be shown by the transport if the transport's
-    camera is enabled. Will play at the framerate specified in the transport's
-    `fps` constructor parameter."""
-    images: list[bytes]
-
-    def __str__(self):
-        return f"{self.__class__.__name__}, list size: {len(self.images)}"
-
-
-@dataclass()
-class TextFrame(Frame):
-    """A chunk of text. Emitted by LLM services, consumed by TTS services, can
-    be used to send text through pipelines."""
-    text: str
-
-    def __str__(self):
-        return f'{self.__class__.__name__}: "{self.text}"'
-
-
-@dataclass()
-class TranscriptionFrame(TextFrame):
-    """A text frame with transcription-specific data. Will be placed in the
-    transport's receive queue when a participant speaks."""
-    participantId: str
-    timestamp: str
-
-    def __str__(self):
-        return f"{self.__class__.__name__}, text: '{self.text}' participantId: {self.participantId}, timestamp: {self.timestamp}"
-
-
-class TTSStartFrame(ControlFrame):
-    """Used to indicate the beginning of a TTS response. Following AudioFrames
-    are part of the TTS response until an TTEndFrame. These frames can be used
-    for aggregating audio frames in a transport to optimize the size of frames
-    sent to the session, without needing to control this in the TTS service."""
-    pass
-
-
-class TTSEndFrame(ControlFrame):
-    """Indicates the end of a TTS response."""
-    pass
-
-
-@dataclass()
-class LLMMessagesFrame(Frame):
-    """A frame containing a list of LLM messages. Used to signal that an LLM
-    service should run a chat completion and emit an LLMStartFrames, TextFrames
-    and an LLMEndFrame.
-    Note that the messages property on this class is mutable, and will be
-    be updated by various ResponseAggregator frame processors."""
-    messages: List[dict]
-
-
-@dataclass()
-class ReceivedAppMessageFrame(Frame):
-    message: Any
-    sender: str
-
-    def __str__(self):
-        return f"ReceivedAppMessageFrame: sender: {self.sender}, message: {self.message}"
-
-
-@dataclass()
-class SendAppMessageFrame(Frame):
-    message: Any
-    participantId: str | None
-
-    def __str__(self):
-        return f"SendAppMessageFrame: participantId: {self.participantId}, message: {self.message}"
-
-
-class UserStartedSpeakingFrame(Frame):
-    """Emitted by VAD to indicate that a participant has started speaking.
-    This can be used for interruptions or other times when detecting that
-    someone is speaking is more important than knowing what they're saying
-    (as you will with a TranscriptionFrame)"""
-    pass
-
-
-class UserStoppedSpeakingFrame(Frame):
-    """Emitted by the VAD to indicate that a user stopped speaking."""
-    pass
-
-
-class BotStartedSpeakingFrame(Frame):
-    pass
-
-
-class BotStoppedSpeakingFrame(Frame):
-    pass
-
-
-@dataclass()
-class LLMFunctionStartFrame(Frame):
-    """Emitted when the LLM receives the beginning of a function call
-    completion. A frame processor can use this frame to indicate that it should
-    start preparing to make a function call, if it can do so in the absence of
-    any arguments."""
-    function_name: str
-
-
-@dataclass()
-class LLMFunctionCallFrame(Frame):
-    """Emitted when the LLM has received an entire function call completion."""
-    function_name: str
-    arguments: str

src/dailyai/pipeline/openai_frames.py

@@ -1,12 +0,0 @@
-from dataclasses import dataclass
-
-from dailyai.pipeline.frames import Frame
-from dailyai.services.openai_llm_context import OpenAILLMContext
-
-
-@dataclass()
-class OpenAILLMContextFrame(Frame):
-    """Like an LLMMessagesFrame, but with extra context specific to the
-    OpenAI API. The context in this message is also mutable, and will be
-    changed by the OpenAIContextAggregator frame processor."""
-    context: OpenAILLMContext

src/dailyai/pipeline/pipeline.py

@@ -1,149 +0,0 @@
-import asyncio
-import logging
-from typing import AsyncGenerator, AsyncIterable, Iterable, List
-from dailyai.pipeline.frame_processor import FrameProcessor
-
-from dailyai.pipeline.frames import AudioFrame, EndPipeFrame, EndFrame, Frame
-
-
-class Pipeline:
-    """
-    This class manages a pipe of FrameProcessors, and runs them in sequence. The "source"
-    and "sink" queues are managed by the caller. You can use this class stand-alone to
-    perform specialized processing, or you can use the Transport's run_pipeline method to
-    instantiate and run a pipeline with the Transport's sink and source queues.
-    """
-
-    def __init__(
-        self,
-        processors: List[FrameProcessor],
-        source: asyncio.Queue | None = None,
-        sink: asyncio.Queue[Frame] | None = None,
-        name: str | None = None,
-    ):
-        """Create a new pipeline. By default we create the sink and source queues
-        if they're not provided, but these can be overridden to point to other
-        queues. If this pipeline is run by a transport, its sink and source queues
-        will be overridden.
-        """
-        self._processors: List[FrameProcessor] = processors
-
-        self.source: asyncio.Queue[Frame] = source or asyncio.Queue()
-        self.sink: asyncio.Queue[Frame] = sink or asyncio.Queue()
-
-        self._logger = logging.getLogger("dailyai.pipeline")
-        self._last_log_line = ""
-        self._shown_repeated_log = False
-        self._name = name or str(id(self))
-
-    def set_source(self, source: asyncio.Queue[Frame]):
-        """Set the source queue for this pipeline. Frames from this queue
-        will be processed by each frame_processor in the pipeline, or order
-        from first to last."""
-        self.source = source
-
-    def set_sink(self, sink: asyncio.Queue[Frame]):
-        """Set the sink queue for this pipeline. After the last frame_processor
-        has processed a frame, its output will be placed on this queue."""
-        self.sink = sink
-
-    def add_processor(self, processor: FrameProcessor):
-        self._processors.append(processor)
-
-    async def get_next_source_frame(self) -> AsyncGenerator[Frame, None]:
-        """Convenience function to get the next frame from the source queue. This
-        lets us consistently have an AsyncGenerator yield frames, from either the
-        source queue or a frame_processor."""
-
-        yield await self.source.get()
-
-    async def queue_frames(
-        self,
-        frames: Iterable[Frame] | AsyncIterable[Frame],
-    ) -> None:
-        """Insert frames directly into a pipeline. This is typically used inside a transport
-        participant_joined callback to prompt a bot to start a conversation, for example."""
-
-        if isinstance(frames, AsyncIterable):
-            async for frame in frames:
-                await self.source.put(frame)
-        elif isinstance(frames, Iterable):
-            for frame in frames:
-                await self.source.put(frame)
-        else:
-            raise Exception("Frames must be an iterable or async iterable")
-
-    async def run_pipeline(self):
-        """Run the pipeline. Take each frame from the source queue, pass it to
-        the first frame_processor, pass the output of that frame_processor to the
-        next in the list, etc. until the last frame_processor has processed the
-        resulting frames, then place those frames in the sink queue.
-
-        The source and sink queues must be set before calling this method.
-
-        This method will exit when an EndFrame is placed on the sink queue.
-        No more frames will be placed on the sink queue after an EndFrame, even
-        if it's not the last frame yielded by the last frame_processor in the pipeline..
-        """
-
-        try:
-            while True:
-                initial_frame = await self.source.get()
-                async for frame in self._run_pipeline_recursively(
-                    initial_frame, self._processors
-                ):
-                    self._log_frame(frame, len(self._processors) + 1)
-                    await self.sink.put(frame)
-
-                if isinstance(initial_frame, EndFrame) or isinstance(
-                    initial_frame, EndPipeFrame
-                ):
-                    break
-        except asyncio.CancelledError:
-            # this means there's been an interruption, do any cleanup necessary
-            # here.
-            for processor in self._processors:
-                await processor.interrupted()
-
-    async def _run_pipeline_recursively(
-        self, initial_frame: Frame, processors: List[FrameProcessor], depth=1
-    ) -> AsyncGenerator[Frame, None]:
-        """Internal function to add frames to the pipeline as they're yielded
-        by each processor."""
-        if processors:
-            self._log_frame(initial_frame, depth)
-            async for frame in processors[0].process_frame(initial_frame):
-                async for final_frame in self._run_pipeline_recursively(
-                    frame, processors[1:], depth + 1
-                ):
-                    yield final_frame
-        else:
-            yield initial_frame
-
-    def _log_frame(self, frame: Frame, depth: int):
-        """Log a frame as it moves through the pipeline. This is useful for debugging.
-        Note that this function inherits the logging level from the "dailyai" logger.
-        If you want debug output from dailyai in general but not this function (it is
-        noisy) you can silence this function by doing something like this:
-
-        # enable debug logging for the dailyai package.
-        logger = logging.getLogger("dailyai")
-        logger.setLevel(logging.DEBUG)
-
-        # silence the pipeline logging
-        logger = logging.getLogger("dailyai.pipeline")
-        logger.setLevel(logging.WARNING)
-        """
-        source = str(self._processors[depth - 2]) if depth > 1 else "source"
-        dest = str(self._processors[depth - 1]) if depth < (len(self._processors) + 1) else "sink"
-        prefix = self._name + "  " * depth
-        logline = prefix + " -> ".join([source, frame.__class__.__name__, dest])
-        if logline == self._last_log_line:
-            if self._shown_repeated_log:
-                return
-            self._shown_repeated_log = True
-            self._logger.debug(prefix + "... repeated")
-        else:
-            self._shown_repeated_log = False
-            self._last_log_line = logline
-            self._logger.debug(logline)

src/dailyai/services/ai_services.py

@@ -1,145 +0,0 @@
-import io
-import logging
-import time
-import wave
-from dailyai.pipeline.frame_processor import FrameProcessor
-
-from dailyai.pipeline.frames import (
-    AudioFrame,
-    EndFrame,
-    EndPipeFrame,
-    ImageFrame,
-    Frame,
-    TTSEndFrame,
-    TTSStartFrame,
-    TextFrame,
-    TranscriptionFrame,
-)
-
-from abc import abstractmethod
-from typing import AsyncGenerator, BinaryIO
-
-
-class AIService(FrameProcessor):
-    def __init__(self):
-        self.logger = logging.getLogger("dailyai")
-
-
-class LLMService(AIService):
-    """This class is a no-op but serves as a base class for LLM services."""
-
-    def __init__(self):
-        super().__init__()
-
-
-class TTSService(AIService):
-    def __init__(self, aggregate_sentences=True):
-        super().__init__()
-        self.aggregate_sentences: bool = aggregate_sentences
-        self.current_sentence: str = ""
-
-    # Some TTS services require a specific sample rate. We default to 16k
-    def get_mic_sample_rate(self):
-        return 16000
-
-    # Converts the text to audio. Yields a list of audio frames that can
-    # be sent to the microphone device
-    @abstractmethod
-    async def run_tts(self, text) -> AsyncGenerator[bytes, None]:
-        # yield empty bytes here, so linting can infer what this method does
-        yield bytes()
-
-    async def wrap_tts(self, text) -> AsyncGenerator[Frame, None]:
-        yield TTSStartFrame()
-        async for audio_chunk in self.run_tts(text):
-            yield AudioFrame(audio_chunk)
-        yield TTSEndFrame()
-        yield TextFrame(text)
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, EndFrame) or isinstance(frame, EndPipeFrame):
-            if self.current_sentence:
-                async for cleanup_frame in self.wrap_tts(self.current_sentence):
-                    yield cleanup_frame
-
-        if not isinstance(frame, TextFrame):
-            yield frame
-            return
-
-        text: str | None = None
-        if not self.aggregate_sentences:
-            text = frame.text
-        else:
-            self.current_sentence += frame.text
-            if self.current_sentence.strip().endswith((".", "?", "!")):
-                text = self.current_sentence
-                self.current_sentence = ""
-
-        if text:
-            async for frame in self.wrap_tts(text):
-                yield frame
-
-
-class ImageGenService(AIService):
-    def __init__(self, image_size, **kwargs):
-        super().__init__(**kwargs)
-        self.image_size = image_size
-
-    # Renders the image. Returns an Image object.
-    @abstractmethod
-    async def run_image_gen(self, sentence: str) -> tuple[str, bytes]:
-        pass
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if not isinstance(frame, TextFrame):
-            yield frame
-            return
-
-        (url, image_data) = await self.run_image_gen(frame.text)
-        yield ImageFrame(url, image_data)
-
-
-class STTService(AIService):
-    """STTService is a base class for speech-to-text services."""
-
-    _frame_rate: int
-
-    def __init__(self, frame_rate: int = 16000, **kwargs):
-        super().__init__(**kwargs)
-        self._frame_rate = frame_rate
-
-    @abstractmethod
-    async def run_stt(self, audio: BinaryIO) -> str:
-        """Returns transcript as a string"""
-        pass
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        """Processes a frame of audio data, either buffering or transcribing it."""
-        if not isinstance(frame, AudioFrame):
-            return
-
-        data = frame.data
-        content = io.BufferedRandom(io.BytesIO())
-        ww = wave.open(self._content, "wb")
-        ww.setnchannels(1)
-        ww.setsampwidth(2)
-        ww.setframerate(self._frame_rate)
-        ww.writeframesraw(data)
-        ww.close()
-        content.seek(0)
-        text = await self.run_stt(content)
-        yield TranscriptionFrame(text, "", str(time.time()))
-
-
-class FrameLogger(AIService):
-    def __init__(self, prefix="Frame", **kwargs):
-        super().__init__(**kwargs)
-        self.prefix = prefix
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, (AudioFrame, ImageFrame)):
-            self.logger.info(f"{self.prefix}: {type(frame)}")
-        else:
-            print(f"{self.prefix}: {frame}")
-
-        yield frame

src/dailyai/services/anthropic_llm_service.py

@@ -1,44 +0,0 @@
-from typing import AsyncGenerator
-from dailyai.pipeline.frames import Frame, LLMMessagesFrame, TextFrame
-
-from dailyai.services.ai_services import LLMService
-
-try:
-    from anthropic import AsyncAnthropic
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use Anthropic, you need to `pip install dailyai[anthropic]`. Also, set `ANTHROPIC_API_KEY` environment variable.")
-    raise Exception(f"Missing module: {e}")
-
-
-class AnthropicLLMService(LLMService):
-
-    def __init__(
-            self,
-            api_key,
-            model="claude-3-opus-20240229",
-            max_tokens=1024):
-        super().__init__()
-        self.client = AsyncAnthropic(api_key=api_key)
-        self.model = model
-        self.max_tokens = max_tokens
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if not isinstance(frame, LLMMessagesFrame):
-            yield frame
-
-        stream = await self.client.messages.create(
-            max_tokens=self.max_tokens,
-            messages=[
-                {
-                    "role": "user",
-                    "content": "Hello, Claude",
-                }
-            ],
-            model=self.model,
-            stream=True,
-        )
-        async for event in stream:
-            if event.type == "content_block_delta":
-                yield TextFrame(event.delta.text)

src/dailyai/services/deepgram_ai_service.py

@@ -1,36 +0,0 @@
-import aiohttp
-
-from dailyai.services.ai_services import TTSService
-
-
-class DeepgramAIService(TTSService):
-    def __init__(
-        self,
-        *,
-        aiohttp_session: aiohttp.ClientSession,
-        api_key,
-        voice,
-        sample_rate=16000
-    ):
-        super().__init__()
-
-        self._api_key = api_key
-        self._voice = voice
-        self._sample_rate = sample_rate
-        self._aiohttp_session = aiohttp_session
-
-    async def run_tts(self, sentence):
-        self.logger.info(f"Running deepgram tts for {sentence}")
-        base_url = "https://api.beta.deepgram.com/v1/speak"
-        request_url = f"{base_url}?model={self._voice}&encoding=linear16&container=none&sample_rate={self._sample_rate}"
-        headers = {
-            "authorization": f"token {self._api_key}",
-            "Content-Type": "application/json"}
-        data = {"text": sentence}
-
-        async with self._aiohttp_session.post(
-            request_url, headers=headers, json=data
-        ) as r:
-            async for chunk in r.content:
-                if chunk:
-                    yield chunk

src/dailyai/services/elevenlabs_ai_service.py

@@ -1,46 +0,0 @@
-import aiohttp
-
-from typing import AsyncGenerator
-
-from dailyai.services.ai_services import TTSService
-
-
-class ElevenLabsTTSService(TTSService):
-
-    def __init__(
-        self,
-        *,
-        aiohttp_session: aiohttp.ClientSession,
-        api_key,
-        voice_id,
-        model="eleven_turbo_v2",
-    ):
-        super().__init__()
-
-        self._api_key = api_key
-        self._voice_id = voice_id
-        self._aiohttp_session = aiohttp_session
-        self._model = model
-
-    async def run_tts(self, sentence) -> AsyncGenerator[bytes, None]:
-        url = f"https://api.elevenlabs.io/v1/text-to-speech/{self._voice_id}/stream"
-        payload = {"text": sentence, "model_id": self._model}
-        querystring = {
-            "output_format": "pcm_16000",
-            "optimize_streaming_latency": 2}
-        headers = {
-            "xi-api-key": self._api_key,
-            "Content-Type": "application/json",
-        }
-        async with self._aiohttp_session.post(
-            url, json=payload, headers=headers, params=querystring
-        ) as r:
-            if r.status != 200:
-                self.logger.error(
-                    f"audio fetch status code: {r.status}, error: {r.text}"
-                )
-                return
-
-            async for chunk in r.content:
-                if chunk:
-                    yield chunk

src/dailyai/services/fal_ai_services.py

@@ -1,58 +0,0 @@
-import aiohttp
-import asyncio
-import io
-import os
-from PIL import Image
-
-from dailyai.services.ai_services import ImageGenService
-
-try:
-    import fal
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use Fal, you need to `pip install dailyai[fal]`. Also, set `FAL_KEY_ID` and `FAL_KEY_SECRET` environment variables.")
-    raise Exception(f"Missing module: {e}")
-
-
-class FalImageGenService(ImageGenService):
-    def __init__(
-        self,
-        *,
-        image_size,
-        aiohttp_session: aiohttp.ClientSession,
-        key_id=None,
-        key_secret=None
-    ):
-        super().__init__(image_size)
-        self._aiohttp_session = aiohttp_session
-        if key_id:
-            os.environ["FAL_KEY_ID"] = key_id
-        if key_secret:
-            os.environ["FAL_KEY_SECRET"] = key_secret
-
-    async def run_image_gen(self, sentence) -> tuple[str, bytes]:
-        def get_image_url(sentence, size):
-            handler = fal.apps.submit(
-                "110602490-fast-sdxl",
-                # "fal-ai/fast-sdxl",
-                arguments={"prompt": sentence},
-            )
-            for event in handler.iter_events():
-                if isinstance(event, fal.apps.InProgress):
-                    pass
-
-            result = handler.get()
-
-            image_url = result["images"][0]["url"] if result else None
-            if not image_url:
-                raise Exception("Image generation failed")
-
-            return image_url
-
-        image_url = await asyncio.to_thread(get_image_url, sentence, self.image_size)
-        # Load the image from the url
-        async with self._aiohttp_session.get(image_url) as response:
-            image_stream = io.BytesIO(await response.content.read())
-            image = Image.open(image_stream)
-            return (image_url, image.tobytes())

src/dailyai/services/local_stt_service.py

@@ -1,74 +0,0 @@
-import array
-import io
-import math
-import time
-from typing import AsyncGenerator
-import wave
-from dailyai.pipeline.frames import AudioFrame, Frame, TranscriptionFrame
-from dailyai.services.ai_services import STTService
-
-
-class LocalSTTService(STTService):
-    _content: io.BufferedRandom
-    _wave: wave.Wave_write
-    _current_silence_frames: int
-
-    # Configuration
-    _min_rms: int
-    _max_silence_frames: int
-    _frame_rate: int
-
-    def __init__(self,
-                 min_rms: int = 400,
-                 max_silence_frames: int = 3,
-                 frame_rate: int = 16000,
-                 **kwargs):
-        super().__init__(frame_rate, **kwargs)
-        self._current_silence_frames = 0
-        self._min_rms = min_rms
-        self._max_silence_frames = max_silence_frames
-        self._frame_rate = frame_rate
-        self._new_wave()
-
-    def _new_wave(self):
-        """Creates a new wave object and content buffer."""
-        self._content = io.BufferedRandom(io.BytesIO())
-        ww = wave.open(self._content, "wb")
-        ww.setnchannels(1)
-        ww.setsampwidth(2)
-        ww.setframerate(self._frame_rate)
-        self._wave = ww
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        """Processes a frame of audio data, either buffering or transcribing it."""
-        if not isinstance(frame, AudioFrame):
-            yield frame
-            return
-
-        data = frame.data
-        # Try to filter out empty background noise
-        # (Very rudimentary approach, can be improved)
-        rms = self._get_volume(data)
-        if rms >= self._min_rms:
-            # If volume is high enough, write new data to wave file
-            self._wave.writeframesraw(data)
-
-        # If buffer is not empty and we detect a 3-frame pause in speech,
-        # transcribe the audio gathered so far.
-        if self._content.tell() > 0 and self._current_silence_frames > self._max_silence_frames:
-            self._current_silence_frames = 0
-            self._wave.close()
-            self._content.seek(0)
-            text = await self.run_stt(self._content)
-            self._new_wave()
-            yield TranscriptionFrame(text, '', str(time.time()))
-        # If we get this far, this is a frame of silence
-        self._current_silence_frames += 1
-
-    def _get_volume(self, audio: bytes) -> float:
-        # https://docs.python.org/3/library/array.html
-        audio_array = array.array('h', audio)
-        squares = [sample**2 for sample in audio_array]
-        mean = sum(squares) / len(audio_array)
-        rms = math.sqrt(mean)
-        return rms

src/dailyai/services/open_ai_services.py

@@ -1,56 +0,0 @@
-import aiohttp
-from PIL import Image
-import io
-
-from dailyai.services.ai_services import ImageGenService
-from dailyai.services.openai_api_llm_service import BaseOpenAILLMService
-
-
-try:
-    from openai import AsyncOpenAI
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use OpenAI, you need to `pip install dailyai[openai]`. Also, set `OPENAI_API_KEY` environment variable.")
-    raise Exception(f"Missing module: {e}")
-
-
-class OpenAILLMService(BaseOpenAILLMService):
-
-    def __init__(self, model="gpt-4", * args, **kwargs):
-        super().__init__(model, *args, **kwargs)
-
-
-class OpenAIImageGenService(ImageGenService):
-
-    def __init__(
-        self,
-        *,
-        image_size: str,
-        aiohttp_session: aiohttp.ClientSession,
-        api_key,
-        model="dall-e-3",
-    ):
-        super().__init__(image_size=image_size)
-        self._model = model
-        self._client = AsyncOpenAI(api_key=api_key)
-        self._aiohttp_session = aiohttp_session
-
-    async def run_image_gen(self, sentence) -> tuple[str, bytes]:
-        self.logger.info("Generating OpenAI image", sentence)
-
-        image = await self._client.images.generate(
-            prompt=sentence,
-            model=self._model,
-            n=1,
-            size=self.image_size
-        )
-        image_url = image.data[0].url
-        if not image_url:
-            raise Exception("No image provided in response", image)
-
-        # Load the image from the url
-        async with self._aiohttp_session.get(image_url) as response:
-            image_stream = io.BytesIO(await response.content.read())
-            image = Image.open(image_stream)
-            return (image_url, image.tobytes())

src/dailyai/services/openai_llm_context.py

@@ -1,61 +0,0 @@
-from typing import List
-
-try:
-    from openai._types import NOT_GIVEN, NotGiven
-
-    from openai.types.chat import (
-        ChatCompletionToolParam,
-        ChatCompletionToolChoiceOptionParam,
-        ChatCompletionMessageParam,
-    )
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use OpenAI, you need to `pip install dailyai[openai]`. Also, set `OPENAI_API_KEY` environment variable.")
-    raise Exception(f"Missing module: {e}")
-
-
-class OpenAILLMContext:
-
-    def __init__(
-        self,
-        messages: List[ChatCompletionMessageParam] | None = None,
-        tools: List[ChatCompletionToolParam] | NotGiven = NOT_GIVEN,
-        tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven = NOT_GIVEN
-    ):
-        self.messages: List[ChatCompletionMessageParam] = messages if messages else [
-        ]
-        self.tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven = tool_choice
-        self.tools: List[ChatCompletionToolParam] | NotGiven = tools
-
-    @staticmethod
-    def from_messages(messages: List[dict]) -> "OpenAILLMContext":
-        context = OpenAILLMContext()
-        for message in messages:
-            context.add_message({
-                "content": message["content"],
-                "role": message["role"],
-                "name": message["name"] if "name" in message else message["role"]
-            })
-        return context
-
-    # def __deepcopy__(self, memo):
-
-    def add_message(self, message: ChatCompletionMessageParam):
-        self.messages.append(message)
-
-    def get_messages(self) -> List[ChatCompletionMessageParam]:
-        return self.messages
-
-    def set_tool_choice(
-        self, tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven
-    ):
-        self.tool_choice = tool_choice
-
-    def set_tools(
-            self,
-            tools: List[ChatCompletionToolParam] | NotGiven = NOT_GIVEN):
-        if tools != NOT_GIVEN and len(tools) == 0:
-            tools = NOT_GIVEN
-
-        self.tools = tools

src/dailyai/transports/abstract_transport.py

@@ -1,41 +0,0 @@
-from abc import abstractmethod
-import asyncio
-import logging
-import time
-
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.pipeline.pipeline import Pipeline
-
-
-class AbstractTransport:
-    def __init__(self, **kwargs):
-        self.send_queue = asyncio.Queue()
-        self.receive_queue = asyncio.Queue()
-        self.completed_queue = asyncio.Queue()
-
-        duration_minutes = kwargs.get("duration_minutes") or 10
-        self._expiration = time.time() + duration_minutes * 60
-
-        self._mic_enabled = kwargs.get("mic_enabled") or False
-        self._mic_sample_rate = kwargs.get("mic_sample_rate") or 16000
-        self._camera_enabled = kwargs.get("camera_enabled") or False
-        self._camera_width = kwargs.get("camera_width") or 1024
-        self._camera_height = kwargs.get("camera_height") or 768
-        self._speaker_enabled = kwargs.get("speaker_enabled") or False
-        self._speaker_sample_rate = kwargs.get("speaker_sample_rate") or 16000
-        self._fps = kwargs.get("fps") or 8
-
-        self._logger: logging.Logger = logging.getLogger("dailyai.transport")
-
-    @abstractmethod
-    async def run(self, pipeline: Pipeline, override_pipeline_source_queue=True):
-        pass
-
-    @abstractmethod
-    async def run_interruptible_pipeline(
-        self,
-        pipeline: Pipeline,
-        pre_processor: FrameProcessor | None = None,
-        post_processor: FrameProcessor | None = None,
-    ):
-        pass

src/dailyai/transports/daily_transport.py

@@ -1,314 +0,0 @@
-import asyncio
-import inspect
-import logging
-import signal
-import threading
-import types
-
-from functools import partial
-from typing import Any
-
-from dailyai.pipeline.frames import (
-    ReceivedAppMessageFrame,
-    TranscriptionFrame,
-)
-
-from threading import Event
-
-try:
-    from daily import (
-        EventHandler,
-        CallClient,
-        Daily,
-        VirtualCameraDevice,
-        VirtualMicrophoneDevice,
-        VirtualSpeakerDevice,
-    )
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use the Daily transport, you need to `pip install dailyai[daily]`.")
-    raise Exception(f"Missing module: {e}")
-
-
-from dailyai.transports.threaded_transport import ThreadedTransport
-
-NUM_CHANNELS = 1
-
-SPEECH_THRESHOLD = 0.90
-VAD_RESET_PERIOD_MS = 2000
-
-
-class DailyTransport(ThreadedTransport, EventHandler):
-    _daily_initialized = False
-    _lock = threading.Lock()
-
-    _speaker_enabled: bool
-    _speaker_sample_rate: int
-    _vad_enabled: bool
-
-    # This is necessary to override EventHandler's __new__ method.
-    def __new__(cls, *args, **kwargs):
-        return super().__new__(cls)
-
-    def __init__(
-        self,
-        room_url: str,
-        token: str | None,
-        bot_name: str,
-        min_others_count: int = 1,
-        start_transcription: bool = False,
-        **kwargs,
-    ):
-        kwargs['has_webrtc_vad'] = True
-        # This will call ThreadedTransport.__init__ method, not EventHandler
-        super().__init__(**kwargs)
-
-        self._room_url: str = room_url
-        self._bot_name: str = bot_name
-        self._token: str | None = token
-        self._min_others_count = min_others_count
-        self._start_transcription = start_transcription
-
-        self._is_interrupted = Event()
-        self._stop_threads = Event()
-
-        self._other_participant_has_joined = False
-        self._my_participant_id = None
-
-        self.transcription_settings = {
-            "language": "en",
-            "tier": "nova",
-            "model": "2-conversationalai",
-            "profanity_filter": True,
-            "redact": False,
-            "extra": {
-                "endpointing": True,
-                "punctuate": False,
-            },
-        }
-
-        self._logger: logging.Logger = logging.getLogger("dailyai")
-
-        self._event_handlers = {}
-
-        self.webrtc_vad = Daily.create_native_vad(
-            reset_period_ms=VAD_RESET_PERIOD_MS,
-            sample_rate=self._speaker_sample_rate,
-            channels=NUM_CHANNELS
-        )
-
-    def _patch_method(self, event_name, *args, **kwargs):
-        try:
-            for handler in self._event_handlers[event_name]:
-                if inspect.iscoroutinefunction(handler):
-                    if self._loop:
-                        future = asyncio.run_coroutine_threadsafe(
-                            handler(*args, **kwargs), self._loop)
-
-                        # wait for the coroutine to finish. This will also
-                        # raise any exceptions raised by the coroutine.
-                        future.result()
-                    else:
-                        raise Exception(
-                            "No event loop to run coroutine. In order to use async event handlers, you must run the DailyTransportService in an asyncio event loop.")
-                else:
-                    handler(*args, **kwargs)
-        except Exception as e:
-            self._logger.error(f"Exception in event handler {event_name}: {e}")
-            raise e
-
-    def _webrtc_vad_analyze(self):
-        buffer = self.read_audio_frames(int(self._vad_samples))
-        if len(buffer) > 0:
-            confidence = self.webrtc_vad.analyze_frames(buffer)
-            # yeses = int(confidence * 20.0)
-            # nos = 20 - yeses
-            # out = "!" * yeses + "." * nos
-            # print(f"!!! confidence: {out} {confidence}")
-            talking = confidence > SPEECH_THRESHOLD
-            return talking
-
-    def add_event_handler(self, event_name: str, handler):
-        if not event_name.startswith("on_"):
-            raise Exception(
-                f"Event handler {event_name} must start with 'on_'")
-
-        methods = inspect.getmembers(self, predicate=inspect.ismethod)
-        if event_name not in [method[0] for method in methods]:
-            raise Exception(f"Event handler {event_name} not found")
-
-        if event_name not in self._event_handlers:
-            self._event_handlers[event_name] = [
-                getattr(
-                    self, event_name), types.MethodType(
-                    handler, self)]
-            setattr(self, event_name, partial(self._patch_method, event_name))
-        else:
-            self._event_handlers[event_name].append(
-                types.MethodType(handler, self))
-
-    def event_handler(self, event_name: str):
-        def decorator(handler):
-            self.add_event_handler(event_name, handler)
-            return handler
-
-        return decorator
-
-    def write_frame_to_camera(self, frame: bytes):
-        self.camera.write_frame(frame)
-
-    def write_frame_to_mic(self, frame: bytes):
-        self.mic.write_frames(frame)
-
-    def send_app_message(self, message: Any, participantId: str | None):
-        self.client.send_app_message(message, participantId)
-
-    def read_audio_frames(self, desired_frame_count):
-        bytes = self._speaker.read_frames(desired_frame_count)
-        return bytes
-
-    def _prerun(self):
-        # Only initialize Daily once
-        if not DailyTransport._daily_initialized:
-            with DailyTransport._lock:
-                Daily.init()
-                DailyTransport._daily_initialized = True
-        self.client = CallClient(event_handler=self)
-
-        if self._mic_enabled:
-            self.mic: VirtualMicrophoneDevice = Daily.create_microphone_device(
-                "mic", sample_rate=self._mic_sample_rate, channels=1
-            )
-
-        if self._camera_enabled:
-            self.camera: VirtualCameraDevice = Daily.create_camera_device(
-                "camera", width=self._camera_width, height=self._camera_height, color_format="RGB")
-
-        if self._speaker_enabled or self._vad_enabled:
-            self._speaker: VirtualSpeakerDevice = Daily.create_speaker_device(
-                "speaker", sample_rate=self._speaker_sample_rate, channels=1
-            )
-            Daily.select_speaker_device("speaker")
-
-        self.client.set_user_name(self._bot_name)
-        self.client.join(
-            self._room_url,
-            self._token,
-            completion=self.call_joined,
-            client_settings={
-                "inputs": {
-                    "camera": {
-                        "isEnabled": True,
-                        "settings": {
-                            "deviceId": "camera",
-                        },
-                    },
-                    "microphone": {
-                        "isEnabled": True,
-                        "settings": {
-                            "deviceId": "mic",
-                            "customConstraints": {
-                                "autoGainControl": {"exact": False},
-                                "echoCancellation": {"exact": False},
-                                "noiseSuppression": {"exact": False},
-                            },
-                        },
-                    },
-                },
-                "publishing": {
-                    "camera": {
-                        "sendSettings": {
-                            "maxQuality": "low",
-                            "encodings": {
-                                "low": {
-                                    "maxBitrate": 250000,
-                                    "scaleResolutionDownBy": 1.333,
-                                    "maxFramerate": 8,
-                                }
-                            },
-                        }
-                    }
-                },
-            },
-        )
-        self._my_participant_id = self.client.participants()["local"]["id"]
-
-        self.client.update_subscription_profiles({
-            "base": {
-                "camera": "unsubscribed",
-            }
-        })
-
-        if self._token and self._start_transcription:
-            self.client.start_transcription(self.transcription_settings)
-
-        self.original_sigint_handler = signal.getsignal(signal.SIGINT)
-        signal.signal(signal.SIGINT, self.process_interrupt_handler)
-
-    def process_interrupt_handler(self, signum, frame):
-        self._post_run()
-        if callable(self.original_sigint_handler):
-            self.original_sigint_handler(signum, frame)
-
-    def _post_run(self):
-        self.client.leave()
-        self.client.release()
-
-    def on_first_other_participant_joined(self):
-        pass
-
-    def call_joined(self, join_data, client_error):
-        # self._logger.info(f"Call_joined: {join_data}, {client_error}")
-        pass
-
-    def dialout(self, number):
-        self.client.start_dialout({"phoneNumber": number})
-
-    def start_recording(self):
-        self.client.start_recording()
-
-    def on_error(self, error):
-        self._logger.error(f"on_error: {error}")
-
-    def on_call_state_updated(self, state):
-        pass
-
-    def on_participant_joined(self, participant):
-        if not self._other_participant_has_joined and participant["id"] != self._my_participant_id:
-            self._other_participant_has_joined = True
-            self.on_first_other_participant_joined()
-
-    def on_participant_left(self, participant, reason):
-        if len(self.client.participants()) < self._min_others_count + 1:
-            self._stop_threads.set()
-
-    def on_app_message(self, message: Any, sender: str):
-        if self._loop:
-            frame = ReceivedAppMessageFrame(message, sender)
-            print(frame)
-            asyncio.run_coroutine_threadsafe(
-                self.receive_queue.put(frame), self._loop
-            )
-
-    def on_transcription_message(self, message: dict):
-        if self._loop:
-            participantId = ""
-            if "participantId" in message:
-                participantId = message["participantId"]
-            elif "session_id" in message:
-                participantId = message["session_id"]
-            if self._my_participant_id and participantId != self._my_participant_id:
-                frame = TranscriptionFrame(
-                    message["text"], participantId, message["timestamp"])
-                asyncio.run_coroutine_threadsafe(
-                    self.receive_queue.put(frame), self._loop)
-
-    def on_transcription_error(self, message):
-        self._logger.error(f"Transcription error: {message}")
-
-    def on_transcription_started(self, status):
-        pass
-
-    def on_transcription_stopped(self, stopped_by, stopped_by_error):
-        pass

src/dailyai/transports/local_transport.py

@@ -1,93 +0,0 @@
-import asyncio
-import numpy as np
-import tkinter as tk
-
-from dailyai.transports.threaded_transport import ThreadedTransport
-
-try:
-    import pyaudio
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use the local transport, you need to `pip install dailyai[local]`. On MacOS, you also need to `brew install portaudio`.")
-    raise Exception(f"Missing module: {e}")
-
-
-class LocalTransport(ThreadedTransport):
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        self._sample_width = kwargs.get("sample_width") or 2
-        self._n_channels = kwargs.get("n_channels") or 1
-        self._tk_root = kwargs.get("tk_root") or None
-
-        if self._camera_enabled and not self._tk_root:
-            raise ValueError(
-                "If camera is enabled, a tkinter root must be provided")
-
-        if self._speaker_enabled:
-            self._speaker_buffer_pending = bytearray()
-
-    async def _write_frame_to_tkinter(self, frame: bytes):
-        data = f"P6 {self._camera_width} {self._camera_height} 255 ".encode() + \
-            frame
-        photo = tk.PhotoImage(
-            width=self._camera_width,
-            height=self._camera_height,
-            data=data,
-            format="PPM")
-        self._image_label.config(image=photo)
-
-        # This holds a reference to the photo, preventing it from being garbage
-        # collected.
-        self._image_label.image = photo  # type: ignore
-
-    def write_frame_to_camera(self, frame: bytes):
-        if self._camera_enabled and self._loop:
-            asyncio.run_coroutine_threadsafe(
-                self._write_frame_to_tkinter(frame), self._loop
-            )
-
-    def write_frame_to_mic(self, frame: bytes):
-        if self._mic_enabled:
-            self._audio_stream.write(frame)
-
-    def read_frames(self, desired_frame_count):
-        bytes = b""
-        if self._speaker_enabled:
-            bytes = self._speaker_stream.read(
-                desired_frame_count,
-                exception_on_overflow=False,
-            )
-        return bytes
-
-    def _prerun(self):
-        if self._mic_enabled:
-            self._pyaudio = pyaudio.PyAudio()
-            self._audio_stream = self._pyaudio.open(
-                format=self._pyaudio.get_format_from_width(self._sample_width),
-                channels=self._n_channels,
-                rate=self._speaker_sample_rate,
-                output=True,
-            )
-
-        if self._camera_enabled:
-            # Start with a neutral gray background.
-            array = np.ones((1024, 1024, 3)) * 128
-            data = f"P5 {1024} {1024} 255 ".encode(
-            ) + array.astype(np.uint8).tobytes()
-            photo = tk.PhotoImage(
-                width=1024,
-                height=1024,
-                data=data,
-                format="PPM")
-            self._image_label = tk.Label(self._tk_root, image=photo)
-            self._image_label.pack()
-
-        if self._speaker_enabled:
-            self._speaker_stream = self._pyaudio.open(
-                format=self._pyaudio.get_format_from_width(self._sample_width),
-                channels=self._n_channels,
-                rate=self._speaker_sample_rate,
-                frames_per_buffer=self._speaker_sample_rate,
-                input=True
-            )

src/dailyai/transports/threaded_transport.py

@@ -1,497 +0,0 @@
-from abc import abstractmethod
-import asyncio
-import itertools
-import numpy as np
-
-import queue
-import threading
-import time
-from typing import Any, AsyncGenerator
-from enum import Enum
-from dailyai.pipeline.frame_processor import FrameProcessor
-
-from dailyai.pipeline.frames import (
-    SendAppMessageFrame,
-    AudioFrame,
-    EndFrame,
-    ImageFrame,
-    Frame,
-    PipelineStartedFrame,
-    SpriteFrame,
-    StartFrame,
-    TextFrame,
-    UserStartedSpeakingFrame,
-    UserStoppedSpeakingFrame,
-)
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.services.ai_services import TTSService
-from dailyai.transports.abstract_transport import AbstractTransport
-
-
-# Provided by Alexander Veysov
-
-
-def int2float(sound):
-    try:
-        abs_max = np.abs(sound).max()
-        sound = sound.astype("float32")
-        if abs_max > 0:
-            sound *= 1 / 32768
-        sound = sound.squeeze()  # depends on the use case
-        return sound
-    except ValueError:
-        return sound
-
-
-class VADState(Enum):
-    QUIET = 1
-    STARTING = 2
-    SPEAKING = 3
-    STOPPING = 4
-
-
-class ThreadedTransport(AbstractTransport):
-
-    def __init__(
-        self,
-        **kwargs,
-    ) -> None:
-        super().__init__(**kwargs)
-        self._vad_start_s = kwargs.get("vad_start_s") or 0.2
-        self._vad_stop_s = kwargs.get("vad_stop_s") or 0.8
-        self._context = kwargs.get("context") or []
-        self._vad_enabled = kwargs.get("vad_enabled") or False
-        self._has_webrtc_vad = kwargs.get("has_webrtc_vad") or False
-        if self._vad_enabled and self._speaker_enabled:
-            raise Exception(
-                "Sorry, you can't use speaker_enabled and vad_enabled at the same time. Please set one to False."
-            )
-        self._vad_samples = 1536
-
-        if self._vad_enabled:
-            try:
-                global torch, torchaudio
-                import torch
-                # We don't use torchaudio here, but we need to try importing it because
-                # Silero uses it
-                import torchaudio
-                torch.set_num_threads(1)
-
-                (self.model, self.utils) = torch.hub.load(
-                    repo_or_dir="snakers4/silero-vad", model="silero_vad", force_reload=False
-                )
-                self._logger.debug("Loaded Silero VAD")
-
-            except ModuleNotFoundError as e:
-                if self._has_webrtc_vad:
-                    self._logger.debug(
-                        f"Couldn't load torch; using webrtc VAD")
-                    self._vad_samples = int(self._speaker_sample_rate / 100.0)
-                else:
-                    self._logger.error(f"Exception: {e}")
-                    self._logger.error(
-                        "In order to use Silero VAD, you'll need to `pip install dailyai[silero].")
-                    raise Exception(f"Missing module(s): {e}")
-
-        vad_frame_s = self._vad_samples / self._speaker_sample_rate
-        self._vad_start_frames = round(self._vad_start_s / vad_frame_s)
-        self._vad_stop_frames = round(self._vad_stop_s / vad_frame_s)
-        self._vad_starting_count = 0
-        self._vad_stopping_count = 0
-        self._vad_state = VADState.QUIET
-        self._user_is_speaking = False
-
-        self._threadsafe_send_queue = queue.Queue()
-
-        self._images = None
-
-        try:
-            self._loop: asyncio.AbstractEventLoop | None = asyncio.get_running_loop()
-        except RuntimeError:
-            self._loop = None
-
-        self._stop_threads = threading.Event()
-        self._is_interrupted = threading.Event()
-
-    async def run(self, pipeline: Pipeline | None = None, override_pipeline_source_queue=True):
-        self._prerun()
-
-        async_output_queue_marshal_task = asyncio.create_task(
-            self._marshal_frames())
-
-        self._camera_thread = threading.Thread(
-            target=self._run_camera, daemon=True)
-        self._camera_thread.start()
-
-        self._frame_consumer_thread = threading.Thread(
-            target=self._frame_consumer, daemon=True
-        )
-        self._frame_consumer_thread.start()
-
-        if self._speaker_enabled:
-            self._receive_audio_thread = threading.Thread(
-                target=self._receive_audio, daemon=True
-            )
-            self._receive_audio_thread.start()
-
-        if self._vad_enabled:
-            self._vad_thread = threading.Thread(target=self._vad, daemon=True)
-            self._vad_thread.start()
-
-        pipeline_task = None
-        if pipeline:
-            pipeline_task = asyncio.create_task(
-                self.run_pipeline(pipeline, override_pipeline_source_queue)
-            )
-
-        try:
-            while time.time() < self._expiration and not self._stop_threads.is_set():
-                await asyncio.sleep(1)
-        except Exception as e:
-            self._logger.error(f"Exception {e}")
-            raise e
-        finally:
-            # Do anything that must be done to clean up
-            self._post_run()
-
-        self._stop_threads.set()
-
-        if pipeline_task:
-            pipeline_task.cancel()
-
-        await self.send_queue.put(EndFrame())
-
-        await async_output_queue_marshal_task
-        self._frame_consumer_thread.join()
-
-        if self._speaker_enabled:
-            self._receive_audio_thread.join()
-
-        if self._vad_enabled:
-            self._vad_thread.join()
-
-    async def run_pipeline(self, pipeline: Pipeline, override_pipeline_source_queue=True):
-        pipeline.set_sink(self.send_queue)
-        if override_pipeline_source_queue:
-            pipeline.set_source(self.receive_queue)
-        await pipeline.run_pipeline()
-
-    async def run_interruptible_pipeline(
-        self,
-        pipeline: Pipeline,
-        pre_processor: FrameProcessor | None = None,
-        post_processor: FrameProcessor | None = None,
-    ):
-        pipeline.set_sink(self.send_queue)
-        source_queue = asyncio.Queue()
-        pipeline.set_source(source_queue)
-        pipeline_task = asyncio.create_task(pipeline.run_pipeline())
-
-        async def yield_frame(frame: Frame) -> AsyncGenerator[Frame, None]:
-            yield frame
-
-        async def post_process(post_processor: FrameProcessor):
-            while True:
-                frame = await self.completed_queue.get()
-
-                # We ignore the output of the post_processor's process frame;
-                # this is called to update the post-processor's state.
-                async for frame in post_processor.process_frame(frame):
-                    pass
-
-                if isinstance(frame, EndFrame):
-                    break
-
-        if post_processor:
-            post_process_task = asyncio.create_task(
-                post_process(post_processor))
-
-        started = False
-
-        async for frame in self.get_receive_frames():
-            if isinstance(frame, UserStartedSpeakingFrame):
-                pipeline_task.cancel()
-                self.interrupt()
-                pipeline_task = asyncio.create_task(pipeline.run_pipeline())
-                started = False
-
-            if not started:
-                await self.send_queue.put(StartFrame())
-
-            if pre_processor:
-                frame_generator = pre_processor.process_frame(frame)
-            else:
-                frame_generator = yield_frame(frame)
-
-            async for frame in frame_generator:
-                await source_queue.put(frame)
-
-            if isinstance(frame, EndFrame):
-                break
-
-        await asyncio.gather(pipeline_task, post_process_task)
-
-    async def say(self, text: str, tts: TTSService):
-        """Say a phrase. Use with caution; this bypasses any running pipelines."""
-        async for frame in tts.process_frame(TextFrame(text)):
-            await self.send_queue.put(frame)
-
-    def _post_run(self):
-        # Note that this function must be idempotent! It can be called multiple times
-        # if, for example, a keyboard interrupt occurs.
-        pass
-
-    def stop(self):
-        self._stop_threads.set()
-
-    async def stop_when_done(self):
-        await self._wait_for_send_queue_to_empty()
-        self.stop()
-
-    async def _wait_for_send_queue_to_empty(self):
-        await self.send_queue.join()
-        self._threadsafe_send_queue.join()
-
-    @abstractmethod
-    def write_frame_to_camera(self, frame: bytes):
-        pass
-
-    @abstractmethod
-    def write_frame_to_mic(self, frame: bytes):
-        pass
-
-    @abstractmethod
-    def read_audio_frames(self, desired_frame_count):
-        return bytes()
-
-    @abstractmethod
-    def _prerun(self):
-        pass
-
-    def _silero_vad_analyze(self):
-        try:
-            audio_chunk = self.read_audio_frames(self._vad_samples)
-            audio_int16 = np.frombuffer(audio_chunk, np.int16)
-            audio_float32 = int2float(audio_int16)
-            new_confidence = self.model(
-                torch.from_numpy(audio_float32), 16000).item()
-            # yeses = int(new_confidence * 20.0)
-            # nos = 20 - yeses
-            # out = "!" * yeses + "." * nos
-            # print(f"!!! confidence: {out}")
-            speaking = new_confidence > 0.5
-            return speaking
-        except BaseException:
-            # This comes from an empty audio array
-            return False
-
-    def _vad(self):
-
-        while not self._stop_threads.is_set():
-            if hasattr(self, 'model'):  # we can use Silero
-                speaking = self._silero_vad_analyze()
-            elif self._has_webrtc_vad:
-                speaking = self._webrtc_vad_analyze()
-            else:
-                raise Exception("VAD is running with no VAD service available")
-            if speaking:
-                match self._vad_state:
-                    case VADState.QUIET:
-                        self._vad_state = VADState.STARTING
-                        self._vad_starting_count = 1
-                    case VADState.STARTING:
-                        self._vad_starting_count += 1
-                    case VADState.STOPPING:
-                        self._vad_state = VADState.SPEAKING
-                        self._vad_stopping_count = 0
-            else:
-                match self._vad_state:
-                    case VADState.STARTING:
-                        self._vad_state = VADState.QUIET
-                        self._vad_starting_count = 0
-                    case VADState.SPEAKING:
-                        self._vad_state = VADState.STOPPING
-                        self._vad_stopping_count = 1
-                    case VADState.STOPPING:
-                        self._vad_stopping_count += 1
-
-            if (
-                self._vad_state == VADState.STARTING
-                and self._vad_starting_count >= self._vad_start_frames
-            ):
-                if self._loop:
-                    asyncio.run_coroutine_threadsafe(
-                        self.receive_queue.put(
-                            UserStartedSpeakingFrame()), self._loop)
-                # self.interrupt()
-                self._vad_state = VADState.SPEAKING
-                self._vad_starting_count = 0
-            if (
-                self._vad_state == VADState.STOPPING
-                and self._vad_stopping_count >= self._vad_stop_frames
-            ):
-
-                if self._loop:
-                    asyncio.run_coroutine_threadsafe(
-                        self.receive_queue.put(
-                            UserStoppedSpeakingFrame()), self._loop)
-                self._vad_state = VADState.QUIET
-                self._vad_stopping_count = 0
-
-    async def _marshal_frames(self):
-        while True:
-            frame: Frame | list = await self.send_queue.get()
-            self._threadsafe_send_queue.put(frame)
-            self.send_queue.task_done()
-            if isinstance(frame, EndFrame):
-                break
-
-    def interrupt(self):
-        self._logger.debug("### Interrupting")
-        self._is_interrupted.set()
-
-    async def get_receive_frames(self) -> AsyncGenerator[Frame, None]:
-        while True:
-            frame = await self.receive_queue.get()
-            yield frame
-            if isinstance(frame, EndFrame):
-                break
-
-    def _receive_audio(self):
-        if not self._loop:
-            self._logger.error("No loop available for audio thread")
-            return
-
-        seconds = 1
-        desired_frame_count = self._speaker_sample_rate * seconds
-        while not self._stop_threads.is_set():
-            buffer = self.read_audio_frames(desired_frame_count)
-            if len(buffer) > 0:
-                frame = AudioFrame(buffer)
-                asyncio.run_coroutine_threadsafe(
-                    self.receive_queue.put(frame), self._loop
-                )
-
-        asyncio.run_coroutine_threadsafe(
-            self.receive_queue.put(
-                EndFrame()), self._loop)
-
-    def _set_image(self, image: bytes):
-        self._images = itertools.cycle([image])
-
-    def _set_images(self, images: list[bytes], start_frame=0):
-        self._images = itertools.cycle(images)
-
-    def send_app_message(self, message: Any, participantId: str | None):
-        """ Child classes should override this to send a custom message to the room. """
-        pass
-
-    def _run_camera(self):
-        try:
-            while not self._stop_threads.is_set():
-                if self._images:
-                    this_frame = next(self._images)
-                    self.write_frame_to_camera(this_frame)
-
-                time.sleep(1.0 / self._fps)
-        except Exception as e:
-            self._logger.error(f"Exception {e} in camera thread.")
-            raise e
-
-    def _frame_consumer(self):
-        self._logger.info("🎬 Starting frame consumer thread")
-        b = bytearray()
-        smallest_write_size = 3200
-        largest_write_size = 8000
-        while True:
-            try:
-                frames_or_frame: Frame | list[Frame] = self._threadsafe_send_queue.get(
-                )
-                if (
-                    isinstance(frames_or_frame, AudioFrame)
-                    and len(frames_or_frame.data) > largest_write_size
-                ):
-                    # subdivide large audio frames to enable interruption
-                    frames = []
-                    for i in range(0, len(frames_or_frame.data),
-                                   largest_write_size):
-                        frames.append(AudioFrame(
-                            frames_or_frame.data[i: i + largest_write_size]))
-                elif isinstance(frames_or_frame, Frame):
-                    frames: list[Frame] = [frames_or_frame]
-                elif isinstance(frames_or_frame, list):
-                    frames: list[Frame] = frames_or_frame
-                else:
-                    raise Exception("Unknown type in output queue")
-
-                for frame in frames:
-                    if isinstance(frame, EndFrame):
-                        self._logger.info("Stopping frame consumer thread")
-                        self._stop_threads.set()
-                        self._threadsafe_send_queue.task_done()
-                        if self._loop:
-                            asyncio.run_coroutine_threadsafe(
-                                self.completed_queue.put(frame), self._loop
-                            )
-                            # Also send the EndFrame to the pipeline so it can stop
-                            asyncio.run_coroutine_threadsafe(
-                                self.receive_queue.put(frame), self._loop
-                            )
-                        return
-
-                    # if interrupted, we just pull frames off the queue and
-                    # discard them
-                    if not self._is_interrupted.is_set():
-                        if frame:
-                            if isinstance(frame, AudioFrame):
-                                chunk = frame.data
-
-                                b.extend(chunk)
-                                truncated_length: int = len(b) - (
-                                    len(b) % smallest_write_size
-                                )
-                                if truncated_length:
-                                    self.write_frame_to_mic(
-                                        bytes(b[:truncated_length]))
-                                    b = b[truncated_length:]
-                            elif isinstance(frame, ImageFrame):
-                                self._set_image(frame.image)
-                            elif isinstance(frame, SpriteFrame):
-                                self._set_images(frame.images)
-                            elif isinstance(frame, SendAppMessageFrame):
-                                self.send_app_message(
-                                    frame.message, frame.participantId)
-                        elif len(b):
-                            self.write_frame_to_mic(bytes(b))
-                            b = bytearray()
-                    else:
-                        # if there are leftover audio bytes, write them now; failing to do so
-                        # can cause static in the audio stream.
-                        if len(b):
-                            truncated_length = len(b) - (len(b) % 160)
-                            self.write_frame_to_mic(
-                                bytes(b[:truncated_length]))
-                            b = bytearray()
-
-                        if isinstance(frame, StartFrame):
-                            self._is_interrupted.clear()
-                            asyncio.run_coroutine_threadsafe(
-                                self.receive_queue.put(PipelineStartedFrame()),
-                                self._loop,
-                            )
-
-                    if self._loop:
-                        asyncio.run_coroutine_threadsafe(
-                            self.completed_queue.put(frame), self._loop
-                        )
-
-                self._threadsafe_send_queue.task_done()
-            except queue.Empty:
-                if len(b):
-                    self.write_frame_to_mic(bytes(b))
-
-                b = bytearray()
-            except Exception as e:
-                self._logger.error(
-                    f"Exception in frame_consumer: {e}, {len(b)}")
-                raise e

src/dailyai/transports/websocket_transport.py

@@ -1,125 +0,0 @@
-import asyncio
-import time
-from typing import AsyncGenerator, List
-
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.pipeline.frames import AudioFrame, ControlFrame, EndFrame, Frame, TTSEndFrame, TTSStartFrame, TextFrame
-from dailyai.pipeline.pipeline import Pipeline
-from dailyai.serializers.protobuf_serializer import ProtobufFrameSerializer
-from dailyai.transports.abstract_transport import AbstractTransport
-from dailyai.transports.threaded_transport import ThreadedTransport
-
-try:
-    import websockets
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use the websocket transport, you need to `pip install dailyai[websocket]`.")
-    raise Exception(f"Missing module: {e}")
-
-
-class WebSocketFrameProcessor(FrameProcessor):
-    """This FrameProcessor filters and mutates frames before they're sent over the websocket.
-    This is necessary to aggregate audio frames into sizes that are cleanly playable by the client"""
-
-    def __init__(
-            self,
-            audio_frame_size: int | None = None,
-            sendable_frames: List[Frame] | None = None):
-        super().__init__()
-        if not audio_frame_size:
-            raise ValueError("audio_frame_size must be provided")
-
-        self._audio_frame_size = audio_frame_size
-        self._sendable_frames = sendable_frames or [TextFrame, AudioFrame]
-        self._audio_buffer = bytes()
-        self._in_tts_audio = False
-
-    async def process_frame(self, frame: Frame) -> AsyncGenerator[Frame, None]:
-        if isinstance(frame, TTSStartFrame):
-            self._in_tts_audio = True
-        elif isinstance(frame, AudioFrame):
-            if self._in_tts_audio:
-                self._audio_buffer += frame.data
-                while len(self._audio_buffer) >= self._audio_frame_size:
-                    yield AudioFrame(self._audio_buffer[:self._audio_frame_size])
-                    self._audio_buffer = self._audio_buffer[self._audio_frame_size:]
-        elif isinstance(frame, TTSEndFrame):
-            self._in_tts_audio = False
-            if self._audio_buffer:
-                yield AudioFrame(self._audio_buffer)
-                self._audio_buffer = bytes()
-        elif type(frame) in self._sendable_frames or isinstance(frame, ControlFrame):
-            yield frame
-
-
-class WebsocketTransport(AbstractTransport):
-    def __init__(self, **kwargs):
-        super().__init__(**kwargs)
-        self._sample_width = kwargs.get("sample_width", 2)
-        self._n_channels = kwargs.get("n_channels", 1)
-        self._port = kwargs.get("port", 8765)
-        self._host = kwargs.get("host", "localhost")
-        self._audio_frame_size = kwargs.get("audio_frame_size", 16000)
-        self._sendable_frames = kwargs.get(
-            "sendable_frames", [
-                TextFrame, AudioFrame, TTSEndFrame, TTSStartFrame])
-        self._serializer = kwargs.get("serializer", ProtobufFrameSerializer())
-
-        self._server: websockets.WebSocketServer | None = None
-        self._websocket: websockets.WebSocketServerProtocol | None = None
-
-        self._connection_handlers = []
-
-    async def run(self, pipeline: Pipeline, override_pipeline_source_queue=True):
-        self._stop_server_event = asyncio.Event()
-        pipeline.set_sink(self.send_queue)
-        if override_pipeline_source_queue:
-            pipeline.set_source(self.receive_queue)
-
-        pipeline.add_processor(WebSocketFrameProcessor(
-            audio_frame_size=self._audio_frame_size,
-            sendable_frames=self._sendable_frames))
-
-        async def timeout():
-            sleep_time = self._expiration - time.time()
-            await asyncio.sleep(sleep_time)
-            self._stop_server_event.set()
-
-        async def send_task():
-            while not self._stop_server_event.is_set():
-                frame = await self.send_queue.get()
-                if isinstance(frame, EndFrame):
-                    self._stop_server_event.set()
-                    break
-                if self._websocket and frame:
-                    proto = self._serializer.serialize(frame)
-                    await self._websocket.send(proto)
-
-        async def start_server():
-            async with websockets.serve(self._websocket_handler, self._host, self._port) as server:
-                self._logger.debug("Websocket server started.")
-                await self._stop_server_event.wait()
-                self._logger.debug("Websocket server stopped.")
-            await self.receive_queue.put(EndFrame())
-
-        timeout_task = asyncio.create_task(timeout())
-        await asyncio.gather(start_server(), send_task(), pipeline.run_pipeline())
-        timeout_task.cancel()
-
-    def on_connection(self, handler):
-        self._connection_handlers.append(handler)
-
-    async def _websocket_handler(self, websocket: websockets.WebSocketServerProtocol, path):
-        if self._websocket:
-            await self._websocket.close()
-            self._logger.warning(
-                "Got another websocket connection; closing first.")
-
-        for handler in self._connection_handlers:
-            await handler()
-
-        self._websocket = websocket
-        async for message in websocket:
-            frame = self._serializer.deserialize(message)
-            await self.receive_queue.put(frame)

src/pipecat/frames/frames.proto

@@ -1,6 +1,12 @@
+//
+// Copyright (c) 2024, Daily
+//
+// SPDX-License-Identifier: BSD 2-Clause License
+//
+
 syntax = "proto3";
 
-package dailyai_proto;
+package pipecat_proto;
 
 message TextFrame {
     string text = 1;

src/pipecat/frames/frames.py

@@ -0,0 +1,276 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from typing import List, Tuple
+
+from dataclasses import dataclass, field
+
+from pipecat.utils.utils import obj_count, obj_id
+
+
+@dataclass
+class Frame:
+    id: int = field(init=False)
+    name: str = field(init=False)
+
+    def __post_init__(self):
+        self.id: int = obj_id()
+        self.name: str = f"{self.__class__.__name__}#{obj_count(self)}"
+
+    def __str__(self):
+        return self.name
+
+
+@dataclass
+class DataFrame(Frame):
+    pass
+
+
+@dataclass
+class AudioRawFrame(DataFrame):
+    """A chunk of audio. Will be played by the transport if the transport's
+    microphone has been enabled.
+
+    """
+    audio: bytes
+    sample_rate: int
+    num_channels: int
+
+    def __post_init__(self):
+        super().__post_init__()
+        self.num_frames = int(len(self.audio) / (self.num_channels * 2))
+
+    def __str__(self):
+        return f"{self.name}(size: {len(self.audio)}, frames: {self.num_frames}, sample_rate: {self.sample_rate}, channels: {self.num_channels})"
+
+
+@dataclass
+class ImageRawFrame(DataFrame):
+    """An image. Will be shown by the transport if the transport's camera is
+    enabled.
+
+    """
+    image: bytes
+    size: Tuple[int, int]
+    format: str
+
+    def __str__(self):
+        return f"{self.name}(size: {self.size}, format: {self.format})"
+
+
+@dataclass
+class URLImageRawFrame(ImageRawFrame):
+    """An image with an associated URL. Will be shown by the transport if the
+    transport's camera is enabled.
+
+    """
+    url: str | None
+
+    def __str__(self):
+        return f"{self.name}(url: {self.url}, size: {self.size}, format: {self.format})"
+
+
+@dataclass
+class VisionImageRawFrame(ImageRawFrame):
+    """An image with an associated text to ask for a description of it. Will be
+    shown by the transport if the transport's camera is enabled.
+
+    """
+    text: str | None
+
+    def __str__(self):
+        return f"{self.name}(text: {self.text}, size: {self.size}, format: {self.format})"
+
+
+@dataclass
+class UserImageRawFrame(ImageRawFrame):
+    """An image associated to a user. Will be shown by the transport if the
+    transport's camera is enabled.
+
+    """
+    user_id: str
+
+    def __str__(self):
+        return f"{self.name}(user: {self.user_id}, size: {self.size}, format: {self.format})"
+
+
+@dataclass
+class SpriteFrame(Frame):
+    """An animated sprite. Will be shown by the transport if the transport's
+    camera is enabled. Will play at the framerate specified in the transport's
+    `fps` constructor parameter.
+
+    """
+    images: List[ImageRawFrame]
+
+    def __str__(self):
+        return f"{self.name}(size: {len(self.images)})"
+
+
+@dataclass
+class TextFrame(DataFrame):
+    """A chunk of text. Emitted by LLM services, consumed by TTS services, can
+    be used to send text through pipelines.
+
+    """
+    text: str
+
+    def __str__(self):
+        return f'{self.name}: "{self.text}"'
+
+
+@dataclass
+class TranscriptionFrame(TextFrame):
+    """A text frame with transcription-specific data. Will be placed in the
+    transport's receive queue when a participant speaks.
+
+    """
+    user_id: str
+    timestamp: str
+
+    def __str__(self):
+        return f"{self.name}(user: {self.user_id}, timestamp: {self.timestamp})"
+
+
+@dataclass
+class InterimTranscriptionFrame(TextFrame):
+    """A text frame with interim transcription-specific data. Will be placed in
+    the transport's receive queue when a participant speaks."""
+    user_id: str
+    timestamp: str
+
+    def __str__(self):
+        return f"{self.name}(user: {self.user_id}, timestamp: {self.timestamp})"
+
+
+@dataclass
+class LLMMessagesFrame(DataFrame):
+    """A frame containing a list of LLM messages. Used to signal that an LLM
+    service should run a chat completion and emit an LLMStartFrames, TextFrames
+    and an LLMEndFrame. Note that the messages property on this class is
+    mutable, and will be be updated by various ResponseAggregator frame
+    processors.
+
+    """
+    messages: List[dict]
+
+#
+# App frames. Application user-defined frames.
+#
+
+
+@dataclass
+class AppFrame(Frame):
+    pass
+
+#
+# System frames
+#
+
+
+@dataclass
+class SystemFrame(Frame):
+    pass
+
+
+@dataclass
+class StartFrame(SystemFrame):
+    """This is the first frame that should be pushed down a pipeline."""
+    pass
+
+
+@dataclass
+class CancelFrame(SystemFrame):
+    """Indicates that a pipeline needs to stop right away."""
+    pass
+
+
+@dataclass
+class ErrorFrame(SystemFrame):
+    """This is used notify upstream that an error has occurred downstream the
+    pipeline."""
+    error: str | None
+
+    def __str__(self):
+        return f"{self.name}(error: {self.error})"
+
+#
+# Control frames
+#
+
+
+@dataclass
+class ControlFrame(Frame):
+    pass
+
+
+@dataclass
+class EndFrame(ControlFrame):
+    """Indicates that a pipeline has ended and frame processors and pipelines
+    should be shut down. If the transport receives this frame, it will stop
+    sending frames to its output channel(s) and close all its threads. Note,
+    that this is a control frame, which means it will received in the order it
+    was sent (unline system frames).
+
+    """
+    pass
+
+
+@dataclass
+class LLMResponseStartFrame(ControlFrame):
+    """Used to indicate the beginning of an LLM response. Following TextFrames
+    are part of the LLM response until an LLMResponseEndFrame"""
+    pass
+
+
+@dataclass
+class LLMResponseEndFrame(ControlFrame):
+    """Indicates the end of an LLM response."""
+    pass
+
+
+@dataclass
+class UserStartedSpeakingFrame(ControlFrame):
+    """Emitted by VAD to indicate that a user has started speaking. This can be
+    used for interruptions or other times when detecting that someone is
+    speaking is more important than knowing what they're saying (as you will
+    with a TranscriptionFrame)
+
+    """
+    pass
+
+
+@dataclass
+class UserStoppedSpeakingFrame(ControlFrame):
+    """Emitted by the VAD to indicate that a user stopped speaking."""
+    pass
+
+
+@dataclass
+class TTSStartedFrame(ControlFrame):
+    """Used to indicate the beginning of a TTS response. Following
+    AudioRawFrames are part of the TTS response until an TTSEndFrame. These
+    frames can be used for aggregating audio frames in a transport to optimize
+    the size of frames sent to the session, without needing to control this in
+    the TTS service.
+
+    """
+    pass
+
+
+@dataclass
+class TTSStoppedFrame(ControlFrame):
+    """Indicates the end of a TTS response."""
+    pass
+
+
+@dataclass
+class UserImageRequestFrame(ControlFrame):
+    """A frame user to request an image from the given user."""
+    user_id: str
+
+    def __str__(self):
+        return f"{self.name}, user: {self.user_id}"

src/pipecat/frames/protobufs/frames_pb2.py

@@ -14,7 +14,7 @@ _sym_db = _symbol_database.Default()
 
 
 
-DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x0c\x66rames.proto\x12\rdailyai_proto\"\x19\n\tTextFrame\x12\x0c\n\x04text\x18\x01 \x01(\t\"\x1a\n\nAudioFrame\x12\x0c\n\x04\x64\x61ta\x18\x01 \x01(\x0c\"L\n\x12TranscriptionFrame\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\x15\n\rparticipantId\x18\x02 \x01(\t\x12\x11\n\ttimestamp\x18\x03 \x01(\t\"\xa2\x01\n\x05\x46rame\x12(\n\x04text\x18\x01 \x01(\x0b\x32\x18.dailyai_proto.TextFrameH\x00\x12*\n\x05\x61udio\x18\x02 \x01(\x0b\x32\x19.dailyai_proto.AudioFrameH\x00\x12:\n\rtranscription\x18\x03 \x01(\x0b\x32!.dailyai_proto.TranscriptionFrameH\x00\x42\x07\n\x05\x66rameb\x06proto3')
+DESCRIPTOR = _descriptor_pool.Default().AddSerializedFile(b'\n\x0c\x66rames.proto\x12\rpipecat_proto\"\x19\n\tTextFrame\x12\x0c\n\x04text\x18\x01 \x01(\t\"\x1a\n\nAudioFrame\x12\x0c\n\x04\x64\x61ta\x18\x01 \x01(\x0c\"L\n\x12TranscriptionFrame\x12\x0c\n\x04text\x18\x01 \x01(\t\x12\x15\n\rparticipantId\x18\x02 \x01(\t\x12\x11\n\ttimestamp\x18\x03 \x01(\t\"\xa2\x01\n\x05\x46rame\x12(\n\x04text\x18\x01 \x01(\x0b\x32\x18.pipecat_proto.TextFrameH\x00\x12*\n\x05\x61udio\x18\x02 \x01(\x0b\x32\x19.pipecat_proto.AudioFrameH\x00\x12:\n\rtranscription\x18\x03 \x01(\x0b\x32!.pipecat_proto.TranscriptionFrameH\x00\x42\x07\n\x05\x66rameb\x06proto3')
 
 _globals = globals()
 _builder.BuildMessageAndEnumDescriptors(DESCRIPTOR, _globals)

src/pipecat/pipeline/merge_pipeline.py

@@ -1,6 +1,6 @@
 from typing import List
-from dailyai.pipeline.frames import EndFrame, EndPipeFrame
-from dailyai.pipeline.pipeline import Pipeline
+from pipecat.pipeline.frames import EndFrame, EndPipeFrame
+from pipecat.pipeline.pipeline import Pipeline
 
 
 class SequentialMergePipeline(Pipeline):

src/pipecat/pipeline/parallel_pipeline.py

@@ -0,0 +1,137 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.frames.frames import CancelFrame, EndFrame, Frame, StartFrame
+
+from loguru import logger
+
+
+class Source(FrameProcessor):
+
+    def __init__(self, upstream_queue: asyncio.Queue):
+        super().__init__()
+        self._up_queue = upstream_queue
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self._up_queue.put(frame)
+            case FrameDirection.DOWNSTREAM:
+                await self.push_frame(frame, direction)
+
+
+class Sink(FrameProcessor):
+
+    def __init__(self, downstream_queue: asyncio.Queue):
+        super().__init__()
+        self._down_queue = downstream_queue
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self.push_frame(frame, direction)
+            case FrameDirection.DOWNSTREAM:
+                await self._down_queue.put(frame)
+
+
+class ParallelPipeline(FrameProcessor):
+    def __init__(self, *args):
+        super().__init__()
+
+        if len(args) == 0:
+            raise Exception(f"ParallelPipeline needs at least one argument")
+
+        self._sources = []
+        self._sinks = []
+
+        self._up_queue = asyncio.Queue()
+        self._down_queue = asyncio.Queue()
+        self._up_task: asyncio.Task | None = None
+        self._down_task: asyncio.Task | None = None
+
+        self._pipelines = []
+
+        logger.debug(f"Creating {self} pipelines")
+        for processors in args:
+            if not isinstance(processors, list):
+                raise TypeError(f"ParallelPipeline argument {processors} is not a list")
+
+            # We add a source at before the pipeline and a sink after.
+            source = Source(self._up_queue)
+            sink = Sink(self._down_queue)
+            self._sources.append(source)
+            self._sinks.append(sink)
+
+            # Create pipeline
+            pipeline = Pipeline(processors)
+            source.link(pipeline)
+            pipeline.link(sink)
+            self._pipelines.append(pipeline)
+
+        logger.debug(f"Finished creating {self} pipelines")
+
+    #
+    # Frame processor
+    #
+
+    async def cleanup(self):
+        await asyncio.gather(*[p.cleanup() for p in self._pipelines])
+
+    async def _start_tasks(self):
+        loop = self.get_event_loop()
+        self._up_task = loop.create_task(self._process_up_queue())
+        self._down_task = loop.create_task(self._process_down_queue())
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, StartFrame):
+            await self._start_tasks()
+
+        if direction == FrameDirection.UPSTREAM:
+            # If we get an upstream frame we process it in each sink.
+            await asyncio.gather(*[s.process_frame(frame, direction) for s in self._sinks])
+        elif direction == FrameDirection.DOWNSTREAM:
+            # If we get a downstream frame we process it in each source.
+            # TODO(aleix): We are creating task for each frame. For real-time
+            # video/audio this might be too slow. We should use an already
+            # created task instead.
+            await asyncio.gather(*[s.process_frame(frame, direction) for s in self._sources])
+
+        # If we get an EndFrame we stop our queue processing tasks and wait on
+        # all the pipelines to finish.
+        if isinstance(frame, CancelFrame) or isinstance(frame, EndFrame):
+            # Use None to indicate when queues should be done processing.
+            await self._up_queue.put(None)
+            await self._down_queue.put(None)
+            if self._up_task:
+                await self._up_task
+            if self._down_task:
+                await self._down_task
+
+    async def _process_up_queue(self):
+        running = True
+        seen_ids = set()
+        while running:
+            frame = await self._up_queue.get()
+            if frame and frame.id not in seen_ids:
+                await self.push_frame(frame, FrameDirection.UPSTREAM)
+                seen_ids.add(frame.id)
+            running = frame is not None
+            self._up_queue.task_done()
+
+    async def _process_down_queue(self):
+        running = True
+        seen_ids = set()
+        while running:
+            frame = await self._down_queue.get()
+            if frame and frame.id not in seen_ids:
+                await self.push_frame(frame, FrameDirection.DOWNSTREAM)
+                seen_ids.add(frame.id)
+            running = frame is not None
+            self._down_queue.task_done()

src/pipecat/pipeline/pipeline.py

@@ -0,0 +1,76 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+
+from typing import Callable, Coroutine, List
+
+from pipecat.frames.frames import Frame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class PipelineSource(FrameProcessor):
+
+    def __init__(self, upstream_push_frame: Callable[[Frame, FrameDirection], Coroutine]):
+        super().__init__()
+        self._upstream_push_frame = upstream_push_frame
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self._upstream_push_frame(frame, direction)
+            case FrameDirection.DOWNSTREAM:
+                await self.push_frame(frame, direction)
+
+
+class PipelineSink(FrameProcessor):
+
+    def __init__(self, downstream_push_frame: Callable[[Frame, FrameDirection], Coroutine]):
+        super().__init__()
+        self._downstream_push_frame = downstream_push_frame
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self.push_frame(frame, direction)
+            case FrameDirection.DOWNSTREAM:
+                await self._downstream_push_frame(frame, direction)
+
+
+class Pipeline(FrameProcessor):
+
+    def __init__(self, processors: List[FrameProcessor]):
+        super().__init__()
+
+        # Add a source and a sink queue so we can forward frames upstream and
+        # downstream outside of the pipeline.
+        self._source = PipelineSource(self.push_frame)
+        self._sink = PipelineSink(self.push_frame)
+        self._processors: List[FrameProcessor] = [self._source] + processors + [self._sink]
+
+        self._link_processors()
+
+    #
+    # Frame processor
+    #
+
+    async def cleanup(self):
+        await self._cleanup_processors()
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if direction == FrameDirection.DOWNSTREAM:
+            await self._source.process_frame(frame, FrameDirection.DOWNSTREAM)
+        elif direction == FrameDirection.UPSTREAM:
+            await self._sink.process_frame(frame, FrameDirection.UPSTREAM)
+
+    async def _cleanup_processors(self):
+        await asyncio.gather(*[p.cleanup() for p in self._processors])
+
+    def _link_processors(self):
+        prev = self._processors[0]
+        for curr in self._processors[1:]:
+            prev.link(curr)
+            prev = curr

src/pipecat/pipeline/runner.py

@@ -0,0 +1,60 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import signal
+
+from pipecat.pipeline.task import PipelineTask
+from pipecat.utils.utils import obj_count, obj_id
+
+from loguru import logger
+
+
+class PipelineRunner:
+
+    def __init__(self, name: str | None = None, handle_sigint: bool = True):
+        self.id: int = obj_id()
+        self.name: str = name or f"{self.__class__.__name__}#{obj_count(self)}"
+        self._loop: asyncio.AbstractEventLoop = asyncio.get_running_loop()
+
+        self._tasks = {}
+        self._running = True
+
+        if handle_sigint:
+            self._setup_sigint()
+
+    async def run(self, task: PipelineTask):
+        logger.debug(f"Runner {self} started running {task}")
+        self._running = True
+        self._tasks[task.name] = task
+        await task.run()
+        del self._tasks[task.name]
+        self._running = False
+        logger.debug(f"Runner {self} finished running {task}")
+
+    async def stop_when_done(self):
+        logger.debug(f"Runner {self} scheduled to stop when all tasks are done")
+        await asyncio.gather(*[t.stop_when_done() for t in self._tasks.values()])
+
+    async def cancel(self):
+        logger.debug(f"Canceling runner {self}")
+        await asyncio.gather(*[t.cancel() for t in self._tasks.values()])
+
+    def is_active(self):
+        return self._running
+
+    def _setup_sigint(self):
+        self._loop.add_signal_handler(
+            signal.SIGINT,
+            lambda *args: asyncio.create_task(self._sigint_handler())
+        )
+
+    async def _sigint_handler(self):
+        logger.warning(f"Ctrl-C detected. Canceling runner {self}")
+        await self.cancel()
+
+    def __str__(self):
+        return self.name

src/pipecat/pipeline/task.py

@@ -0,0 +1,93 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+
+from typing import AsyncIterable, Iterable
+
+from pipecat.frames.frames import CancelFrame, EndFrame, ErrorFrame, Frame, StartFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.utils.utils import obj_count, obj_id
+
+from loguru import logger
+
+
+class Source(FrameProcessor):
+
+    def __init__(self, up_queue: asyncio.Queue):
+        super().__init__()
+        self._up_queue = up_queue
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self._up_queue.put(frame)
+            case FrameDirection.DOWNSTREAM:
+                await self.push_frame(frame, direction)
+
+
+class PipelineTask:
+
+    def __init__(self, pipeline: FrameProcessor):
+        self.id: int = obj_id()
+        self.name: str = f"{self.__class__.__name__}#{obj_count(self)}"
+
+        self._pipeline = pipeline
+
+        self._task_queue = asyncio.Queue()
+        self._up_queue = asyncio.Queue()
+
+        self._source = Source(self._up_queue)
+        self._source.link(pipeline)
+
+    async def stop_when_done(self):
+        logger.debug(f"Task {self} scheduled to stop when done")
+        await self.queue_frame(EndFrame())
+
+    async def cancel(self):
+        logger.debug(f"Canceling pipeline task {self}")
+        await self.queue_frame(CancelFrame())
+
+    async def run(self):
+        await asyncio.gather(self._process_task_queue(), self._process_up_queue())
+
+    async def queue_frame(self, frame: Frame):
+        await self._task_queue.put(frame)
+
+    async def queue_frames(self, frames: Iterable[Frame] | AsyncIterable[Frame]):
+        if isinstance(frames, AsyncIterable):
+            async for frame in frames:
+                await self.queue_frame(frame)
+        elif isinstance(frames, Iterable):
+            for frame in frames:
+                await self.queue_frame(frame)
+        else:
+            raise Exception("Frames must be an iterable or async iterable")
+
+    async def _process_task_queue(self):
+        await self._source.process_frame(StartFrame(), FrameDirection.DOWNSTREAM)
+        running = True
+        while running:
+            frame = await self._task_queue.get()
+            await self._source.process_frame(frame, FrameDirection.DOWNSTREAM)
+            self._task_queue.task_done()
+            running = not (isinstance(frame, CancelFrame) or isinstance(frame, EndFrame))
+        # We just enqueue None to terminate the task.
+        await self._up_queue.put(None)
+
+    async def _process_up_queue(self):
+        running = True
+        while running:
+            frame = await self._up_queue.get()
+            if frame:
+                if isinstance(frame, ErrorFrame):
+                    logger.error(f"Error running app: {frame.error}")
+                    await self.queue_frame(CancelFrame())
+            self._up_queue.task_done()
+            running = frame is not None
+
+    def __str__(self):
+        return self.name

src/pipecat/processors/aggregators/gated.py

@@ -0,0 +1,72 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from typing import List
+
+from pipecat.frames.frames import Frame, SystemFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+from loguru import logger
+
+
+class GatedAggregator(FrameProcessor):
+    """Accumulate frames, with custom functions to start and stop accumulation.
+    Yields gate-opening frame before any accumulated frames, then ensuing frames
+    until and not including the gate-closed frame.
+
+    >>> from pipecat.pipeline.frames import ImageFrame
+
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         if isinstance(frame, TextFrame):
+    ...             print(frame.text)
+    ...         else:
+    ...             print(frame.__class__.__name__)
+
+    >>> aggregator = GatedAggregator(
+    ...     gate_close_fn=lambda x: isinstance(x, LLMResponseStartFrame),
+    ...     gate_open_fn=lambda x: isinstance(x, ImageFrame),
+    ...     start_open=False)
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello")))
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello again.")))
+    >>> asyncio.run(print_frames(aggregator, ImageFrame(image=bytes([]), size=(0, 0))))
+    ImageFrame
+    Hello
+    Hello again.
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Goodbye.")))
+    Goodbye.
+    """
+
+    def __init__(self, gate_open_fn, gate_close_fn, start_open):
+        super().__init__()
+        self._gate_open_fn = gate_open_fn
+        self._gate_close_fn = gate_close_fn
+        self._gate_open = start_open
+        self._accumulator: List[Frame] = []
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        # We must not block system frames.
+        if isinstance(frame, SystemFrame):
+            await self.push_frame(frame, direction)
+            return
+
+        old_state = self._gate_open
+        if self._gate_open:
+            self._gate_open = not self._gate_close_fn(frame)
+        else:
+            self._gate_open = self._gate_open_fn(frame)
+
+        if old_state != self._gate_open:
+            state = "open" if self._gate_open else "closed"
+            logger.debug(f"Gate is now {state} because of {frame}")
+
+        if self._gate_open:
+            await self.push_frame(frame, direction)
+            for frame in self._accumulator:
+                await self.push_frame(frame, direction)
+            self._accumulator = []
+        else:
+            self._accumulator.append(frame)

src/pipecat/processors/aggregators/llm_context.py

@@ -0,0 +1,82 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.frames.frames import Frame, InterimTranscriptionFrame, LLMMessagesFrame, TextFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class LLMContextAggregator(FrameProcessor):
+    def __init__(
+        self,
+        messages: list[dict],
+        role: str,
+        complete_sentences=True,
+        pass_through=True,
+    ):
+        super().__init__()
+        self._messages = messages
+        self._role = role
+        self._sentence = ""
+        self._complete_sentences = complete_sentences
+        self._pass_through = pass_through
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        # We don't do anything with non-text frames, pass it along to next in
+        # the pipeline.
+        if not isinstance(frame, TextFrame):
+            await self.push_frame(frame, direction)
+            return
+
+        # If we get interim results, we ignore them.
+        if isinstance(frame, InterimTranscriptionFrame):
+            return
+
+        # The common case for "pass through" is receiving frames from the LLM that we'll
+        # use to update the "assistant" LLM messages, but also passing the text frames
+        # along to a TTS service to be spoken to the user.
+        if self._pass_through:
+            await self.push_frame(frame, direction)
+
+        # TODO: split up transcription by participant
+        if self._complete_sentences:
+            # type: ignore -- the linter thinks this isn't a TextFrame, even
+            # though we check it above
+            self._sentence += frame.text
+            if self._sentence.endswith((".", "?", "!")):
+                self._messages.append(
+                    {"role": self._role, "content": self._sentence})
+                self._sentence = ""
+                await self.push_frame(LLMMessagesFrame(self._messages))
+        else:
+            # type: ignore -- the linter thinks this isn't a TextFrame, even
+            # though we check it above
+            self._messages.append({"role": self._role, "content": frame.text})
+            await self.push_frame(LLMMessagesFrame(self._messages))
+
+
+class LLMUserContextAggregator(LLMContextAggregator):
+    def __init__(
+            self,
+            messages: list[dict],
+            complete_sentences=True):
+        super().__init__(
+            messages,
+            "user",
+            complete_sentences,
+            pass_through=False)
+
+
+class LLMAssistantContextAggregator(LLMContextAggregator):
+    def __init__(
+            self,
+            messages: list[dict],
+            complete_sentences=True):
+        super().__init__(
+            messages,
+            "assistant",
+            complete_sentences,
+            pass_through=True,
+        )

src/pipecat/processors/aggregators/llm_response.py

@@ -0,0 +1,190 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.frames.frames import (
+    Frame,
+    InterimTranscriptionFrame,
+    LLMMessagesFrame,
+    LLMResponseStartFrame,
+    TextFrame,
+    LLMResponseEndFrame,
+    TranscriptionFrame,
+    UserStartedSpeakingFrame,
+    UserStoppedSpeakingFrame)
+
+
+class LLMResponseAggregator(FrameProcessor):
+
+    def __init__(
+        self,
+        *,
+        messages: list[dict] | None,
+        role: str,
+        start_frame,
+        end_frame,
+        accumulator_frame: TextFrame,
+        interim_accumulator_frame: TextFrame | None = None
+    ):
+        super().__init__()
+
+        self._messages = messages
+        self._role = role
+        self._start_frame = start_frame
+        self._end_frame = end_frame
+        self._accumulator_frame = accumulator_frame
+        self._interim_accumulator_frame = interim_accumulator_frame
+        self._seen_start_frame = False
+        self._seen_end_frame = False
+        self._seen_interim_results = False
+
+        self._aggregation = ""
+        self._aggregating = False
+
+    #
+    # Frame processor
+    #
+
+    # Use cases implemented:
+    #
+    # S: Start, E: End, T: Transcription, I: Interim, X: Text
+    #
+    #        S E -> None
+    #      S T E -> X
+    #    S I T E -> X
+    #    S I E T -> X
+    #  S I E I T -> X
+    #
+    # The following case would not be supported:
+    #
+    #    S I E T1 I T2 -> X
+    #
+    # and T2 would be dropped.
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if not self._messages:
+            return
+
+        send_aggregation = False
+
+        if isinstance(frame, self._start_frame):
+            self._seen_start_frame = True
+            self._aggregating = True
+        elif isinstance(frame, self._end_frame):
+            self._seen_end_frame = True
+
+            # We might have received the end frame but we might still be
+            # aggregating (i.e. we have seen interim results but not the final
+            # text).
+            self._aggregating = self._seen_interim_results
+
+            # Send the aggregation if we are not aggregating anymore (i.e. no
+            # more interim results received).
+            send_aggregation = not self._aggregating
+        elif isinstance(frame, self._accumulator_frame):
+            if self._aggregating:
+                self._aggregation += f" {frame.text}"
+                # We have recevied a complete sentence, so if we have seen the
+                # end frame and we were still aggregating, it means we should
+                # send the aggregation.
+                send_aggregation = self._seen_end_frame
+
+            # We just got our final result, so let's reset interim results.
+            self._seen_interim_results = False
+        elif self._interim_accumulator_frame and isinstance(frame, self._interim_accumulator_frame):
+            self._seen_interim_results = True
+        else:
+            await self.push_frame(frame, direction)
+
+        if send_aggregation:
+            await self._push_aggregation()
+
+    async def _push_aggregation(self):
+        if len(self._aggregation) > 0:
+            self._messages.append({"role": self._role, "content": self._aggregation})
+            frame = LLMMessagesFrame(self._messages)
+            await self.push_frame(frame)
+
+            # Reset
+            self._aggregation = ""
+            self._seen_start_frame = False
+            self._seen_end_frame = False
+            self._seen_interim_results = False
+
+
+class LLMAssistantResponseAggregator(LLMResponseAggregator):
+    def __init__(self, messages: list[dict]):
+        super().__init__(
+            messages=messages,
+            role="assistant",
+            start_frame=LLMResponseStartFrame,
+            end_frame=LLMResponseEndFrame,
+            accumulator_frame=TextFrame
+        )
+
+
+class LLMUserResponseAggregator(LLMResponseAggregator):
+    def __init__(self, messages: list[dict]):
+        super().__init__(
+            messages=messages,
+            role="user",
+            start_frame=UserStartedSpeakingFrame,
+            end_frame=UserStoppedSpeakingFrame,
+            accumulator_frame=TranscriptionFrame,
+            interim_accumulator_frame=InterimTranscriptionFrame
+        )
+
+
+class LLMFullResponseAggregator(FrameProcessor):
+    """This class aggregates Text frames until it receives a
+    LLMResponseEndFrame, then emits the concatenated text as
+    a single text frame.
+
+    given the following frames:
+
+        TextFrame("Hello,")
+        TextFrame(" world.")
+        TextFrame(" I am")
+        TextFrame(" an LLM.")
+        LLMResponseEndFrame()]
+
+    this processor will yield nothing for the first 4 frames, then
+
+        TextFrame("Hello, world. I am an LLM.")
+        LLMResponseEndFrame()
+
+    when passed the last frame.
+
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         if isinstance(frame, TextFrame):
+    ...             print(frame.text)
+    ...         else:
+    ...             print(frame.__class__.__name__)
+
+    >>> aggregator = LLMFullResponseAggregator()
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello,")))
+    >>> asyncio.run(print_frames(aggregator, TextFrame(" world.")))
+    >>> asyncio.run(print_frames(aggregator, TextFrame(" I am")))
+    >>> asyncio.run(print_frames(aggregator, TextFrame(" an LLM.")))
+    >>> asyncio.run(print_frames(aggregator, LLMResponseEndFrame()))
+    Hello, world. I am an LLM.
+    LLMResponseEndFrame
+    """
+
+    def __init__(self):
+        super().__init__()
+        self._aggregation = ""
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            self._aggregation += frame.text
+        elif isinstance(frame, LLMResponseEndFrame):
+            await self.push_frame(TextFrame(self._aggregation))
+            await self.push_frame(frame)
+            self._aggregation = ""
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/processors/aggregators/openai_llm_context.py

@@ -1,6 +1,14 @@
-from typing import AsyncGenerator, Callable
-from dailyai.pipeline.frame_processor import FrameProcessor
-from dailyai.pipeline.frames import (
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from dataclasses import dataclass
+
+from typing import AsyncGenerator, Callable, List
+
+from pipecat.frames.frames import (
     Frame,
     LLMResponseEndFrame,
     LLMResponseStartFrame,
@@ -9,16 +17,58 @@ from dailyai.pipeline.frames import (
     UserStartedSpeakingFrame,
     UserStoppedSpeakingFrame,
 )
-from dailyai.pipeline.openai_frames import OpenAILLMContextFrame
-from dailyai.services.openai_llm_context import OpenAILLMContext
+from pipecat.processors.frame_processor import FrameProcessor
 
-try:
-    from openai.types.chat import ChatCompletionRole
-except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use OpenAI, you need to `pip install dailyai[openai]`. Also, set `OPENAI_API_KEY` environment variable.")
-    raise Exception(f"Missing module: {e}")
+from openai._types import NOT_GIVEN, NotGiven
+
+from openai.types.chat import (
+    ChatCompletionRole,
+    ChatCompletionToolParam,
+    ChatCompletionToolChoiceOptionParam,
+    ChatCompletionMessageParam
+)
+
+
+class OpenAILLMContext:
+
+    def __init__(
+        self,
+        messages: List[ChatCompletionMessageParam] | None = None,
+        tools: List[ChatCompletionToolParam] | NotGiven = NOT_GIVEN,
+        tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven = NOT_GIVEN
+    ):
+        self.messages: List[ChatCompletionMessageParam] = messages if messages else [
+        ]
+        self.tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven = tool_choice
+        self.tools: List[ChatCompletionToolParam] | NotGiven = tools
+
+    @ staticmethod
+    def from_messages(messages: List[dict]) -> "OpenAILLMContext":
+        context = OpenAILLMContext()
+        for message in messages:
+            context.add_message({
+                "content": message["content"],
+                "role": message["role"],
+                "name": message["name"] if "name" in message else message["role"]
+            })
+        return context
+
+    def add_message(self, message: ChatCompletionMessageParam):
+        self.messages.append(message)
+
+    def get_messages(self) -> List[ChatCompletionMessageParam]:
+        return self.messages
+
+    def set_tool_choice(
+        self, tool_choice: ChatCompletionToolChoiceOptionParam | NotGiven
+    ):
+        self.tool_choice = tool_choice
+
+    def set_tools(self, tools: List[ChatCompletionToolParam] | NotGiven = NOT_GIVEN):
+        if tools != NOT_GIVEN and len(tools) == 0:
+            tools = NOT_GIVEN
+
+        self.tools = tools
 
 
 class OpenAIContextAggregator(FrameProcessor):
@@ -113,3 +163,13 @@ class OpenAIAssistantContextAggregator(OpenAIContextAggregator):
             accumulator_frame=TextFrame,
             pass_through=True,
         )
+
+
+@dataclass
+class OpenAILLMContextFrame(Frame):
+    """Like an LLMMessagesFrame, but with extra context specific to the OpenAI
+    API. The context in this message is also mutable, and will be changed by the
+    OpenAIContextAggregator frame processor.
+
+    """
+    context: OpenAILLMContext

src/pipecat/processors/aggregators/parallel_task.py

@@ -0,0 +1,104 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+
+from typing import List
+
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.frames.frames import Frame
+
+from loguru import logger
+
+
+class Source(FrameProcessor):
+
+    def __init__(self, upstream_queue: asyncio.Queue):
+        super().__init__()
+        self._up_queue = upstream_queue
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self._up_queue.put(frame)
+            case FrameDirection.DOWNSTREAM:
+                await self.push_frame(frame, direction)
+
+
+class Sink(FrameProcessor):
+
+    def __init__(self, downstream_queue: asyncio.Queue):
+        super().__init__()
+        self._down_queue = downstream_queue
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                await self.push_frame(frame, direction)
+            case FrameDirection.DOWNSTREAM:
+                await self._down_queue.put(frame)
+
+
+class ParallelTask(FrameProcessor):
+    def __init__(self, *args):
+        super().__init__()
+
+        if len(args) == 0:
+            raise Exception(f"ParallelTask needs at least one argument")
+
+        self._sinks = []
+        self._pipelines = []
+
+        self._up_queue = asyncio.Queue()
+        self._down_queue = asyncio.Queue()
+
+        logger.debug(f"Creating {self} pipelines")
+        for processors in args:
+            if not isinstance(processors, list):
+                raise TypeError(f"ParallelTask argument {processors} is not a list")
+
+            # We add a source at the beginning of the pipeline and a sink at the end.
+            source = Source(self._up_queue)
+            sink = Sink(self._down_queue)
+            processors: List[FrameProcessor] = [source] + processors
+            processors.append(sink)
+
+            # Keep track of sinks. We access the source through the pipeline.
+            self._sinks.append(sink)
+
+            # Create pipeline
+            pipeline = Pipeline(processors)
+            self._pipelines.append(pipeline)
+        logger.debug(f"Finished creating {self} pipelines")
+
+    #
+    # Frame processor
+    #
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if direction == FrameDirection.UPSTREAM:
+            # If we get an upstream frame we process it in each sink.
+            await asyncio.gather(*[s.process_frame(frame, direction) for s in self._sinks])
+        elif direction == FrameDirection.DOWNSTREAM:
+            # If we get a downstream frame we process it in each source (using the pipeline).
+            await asyncio.gather(*[p.process_frame(frame, direction) for p in self._pipelines])
+
+        seen_ids = set()
+        while not self._up_queue.empty():
+            frame = await self._up_queue.get()
+            if frame and frame.id not in seen_ids:
+                await self.push_frame(frame, FrameDirection.UPSTREAM)
+                seen_ids.add(frame.id)
+            self._up_queue.task_done()
+
+        seen_ids = set()
+        while not self._down_queue.empty():
+            frame = await self._down_queue.get()
+            if frame and frame.id not in seen_ids:
+                await self.push_frame(frame, FrameDirection.DOWNSTREAM)
+                seen_ids.add(frame.id)
+            self._down_queue.task_done()

src/pipecat/processors/aggregators/sentence.py

@@ -0,0 +1,50 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import re
+
+from typing import List
+
+from pipecat.frames.frames import EndFrame, Frame, TextFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class SentenceAggregator(FrameProcessor):
+    """This frame processor aggregates text frames into complete sentences.
+
+    Frame input/output:
+        TextFrame("Hello,") -> None
+        TextFrame(" world.") -> TextFrame("Hello world.")
+
+    Doctest:
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         print(frame.text)
+
+    >>> aggregator = SentenceAggregator()
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello,")))
+    >>> asyncio.run(print_frames(aggregator, TextFrame(" world.")))
+    Hello, world.
+    """
+
+    def __init__(self):
+        super().__init__()
+        self._aggregation = ""
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            m = re.search("(.*[?.!])(.*)", frame.text)
+            if m:
+                await self.push_frame(TextFrame(self._aggregation + m.group(1)))
+                self._aggregation = m.group(2)
+            else:
+                self._aggregation += frame.text
+        elif isinstance(frame, EndFrame):
+            if self._aggregation:
+                await self.push_frame(TextFrame(self._aggregation))
+            await self.push_frame(frame)
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/processors/aggregators/user_response.py

@@ -0,0 +1,139 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.frames.frames import (
+    Frame,
+    InterimTranscriptionFrame,
+    TextFrame,
+    TranscriptionFrame,
+    UserStartedSpeakingFrame,
+    UserStoppedSpeakingFrame)
+
+
+class ResponseAggregator(FrameProcessor):
+    """This frame processor aggregates frames between a start and an end frame
+    into complete text frame sentences.
+
+    For example, frame input/output:
+        UserStartedSpeakingFrame() -> None
+        TranscriptionFrame("Hello,") -> None
+        TranscriptionFrame(" world.") -> None
+        UserStoppedSpeakingFrame() -> TextFrame("Hello world.")
+
+    Doctest:
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         if isinstance(frame, TextFrame):
+    ...             print(frame.text)
+
+    >>> aggregator = ResponseAggregator(start_frame = UserStartedSpeakingFrame,
+    ...                                 end_frame=UserStoppedSpeakingFrame,
+    ...                                 accumulator_frame=TranscriptionFrame,
+    ...                                 pass_through=False)
+    >>> asyncio.run(print_frames(aggregator, UserStartedSpeakingFrame()))
+    >>> asyncio.run(print_frames(aggregator, TranscriptionFrame("Hello,", 1, 1)))
+    >>> asyncio.run(print_frames(aggregator, TranscriptionFrame("world.",  1, 2)))
+    >>> asyncio.run(print_frames(aggregator, UserStoppedSpeakingFrame()))
+    Hello, world.
+
+    """
+
+    def __init__(
+        self,
+        *,
+        start_frame,
+        end_frame,
+        accumulator_frame: TextFrame,
+        interim_accumulator_frame: TextFrame | None = None
+    ):
+        super().__init__()
+
+        self._start_frame = start_frame
+        self._end_frame = end_frame
+        self._accumulator_frame = accumulator_frame
+        self._interim_accumulator_frame = interim_accumulator_frame
+        self._seen_start_frame = False
+        self._seen_end_frame = False
+        self._seen_interim_results = False
+
+        self._aggregation = ""
+        self._aggregating = False
+
+    #
+    # Frame processor
+    #
+
+    # Use cases implemented:
+    #
+    # S: Start, E: End, T: Transcription, I: Interim, X: Text
+    #
+    #        S E -> None
+    #      S T E -> X
+    #    S I T E -> X
+    #    S I E T -> X
+    #  S I E I T -> X
+    #
+    # The following case would not be supported:
+    #
+    #    S I E T1 I T2 -> X
+    #
+    # and T2 would be dropped.
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        send_aggregation = False
+
+        if isinstance(frame, self._start_frame):
+            self._seen_start_frame = True
+            self._aggregating = True
+        elif isinstance(frame, self._end_frame):
+            self._seen_end_frame = True
+
+            # We might have received the end frame but we might still be
+            # aggregating (i.e. we have seen interim results but not the final
+            # text).
+            self._aggregating = self._seen_interim_results
+
+            # Send the aggregation if we are not aggregating anymore (i.e. no
+            # more interim results received).
+            send_aggregation = not self._aggregating
+        elif isinstance(frame, self._accumulator_frame):
+            if self._aggregating:
+                self._aggregation += f" {frame.text}"
+                # We have recevied a complete sentence, so if we have seen the
+                # end frame and we were still aggregating, it means we should
+                # send the aggregation.
+                send_aggregation = self._seen_end_frame
+
+            # We just got our final result, so let's reset interim results.
+            self._seen_interim_results = False
+        elif self._interim_accumulator_frame and isinstance(frame, self._interim_accumulator_frame):
+            self._seen_interim_results = True
+        else:
+            await self.push_frame(frame, direction)
+
+        if send_aggregation:
+            await self._push_aggregation()
+
+    async def _push_aggregation(self):
+        if len(self._aggregation) > 0:
+            await self.push_frame(TextFrame(self._aggregation.strip()))
+
+            # Reset
+            self._aggregation = ""
+            self._seen_start_frame = False
+            self._seen_end_frame = False
+            self._seen_interim_results = False
+
+
+class UserResponseAggregator(ResponseAggregator):
+    def __init__(self):
+        super().__init__(
+            start_frame=UserStartedSpeakingFrame,
+            end_frame=UserStoppedSpeakingFrame,
+            accumulator_frame=TranscriptionFrame,
+            interim_accumulator_frame=InterimTranscriptionFrame,
+        )

src/pipecat/processors/aggregators/vision_image_frame.py

@@ -0,0 +1,45 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.frames.frames import Frame, ImageRawFrame, TextFrame, VisionImageRawFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class VisionImageFrameAggregator(FrameProcessor):
+    """This aggregator waits for a consecutive TextFrame and an
+    ImageFrame. After the ImageFrame arrives it will output a VisionImageFrame.
+
+    >>> from pipecat.pipeline.frames import ImageFrame
+
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         print(frame)
+
+    >>> aggregator = VisionImageFrameAggregator()
+    >>> asyncio.run(print_frames(aggregator, TextFrame("What do you see?")))
+    >>> asyncio.run(print_frames(aggregator, ImageFrame(image=bytes([]), size=(0, 0))))
+    VisionImageFrame, text: What do you see?, image size: 0x0, buffer size: 0 B
+
+    """
+
+    def __init__(self):
+        super().__init__()
+        self._describe_text = None
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            self._describe_text = frame.text
+        elif isinstance(frame, ImageRawFrame):
+            if self._describe_text:
+                frame = VisionImageRawFrame(
+                    text=self._describe_text,
+                    image=frame.image,
+                    size=frame.size,
+                    format=frame.format)
+                await self.push_frame(frame)
+                self._describe_text = None
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/processors/filter.py

@@ -0,0 +1,34 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from typing import List
+
+from pipecat.frames.frames import AppFrame, ControlFrame, Frame, SystemFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class Filter(FrameProcessor):
+
+    def __init__(self, types: List[type]):
+        super().__init__()
+        self._types = types
+
+    #
+    # Frame processor
+    #
+
+    def _should_passthrough_frame(self, frame):
+        for t in self._types:
+            if isinstance(frame, t):
+                return True
+
+        return (isinstance(frame, AppFrame)
+                or isinstance(frame, ControlFrame)
+                or isinstance(frame, SystemFrame))
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if self._should_passthrough_frame(frame):
+            await self.push_frame(frame, direction)

src/pipecat/processors/frame_processor.py

@@ -0,0 +1,54 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+from asyncio import AbstractEventLoop
+from enum import Enum
+
+from pipecat.frames.frames import Frame
+from pipecat.utils.utils import obj_count, obj_id
+
+from loguru import logger
+
+
+class FrameDirection(Enum):
+    DOWNSTREAM = 1
+    UPSTREAM = 2
+
+
+class FrameProcessor:
+
+    def __init__(self):
+        self.id: int = obj_id()
+        self.name = f"{self.__class__.__name__}#{obj_count(self)}"
+        self._prev: "FrameProcessor" | None = None
+        self._next: "FrameProcessor" | None = None
+        self._loop: AbstractEventLoop = asyncio.get_running_loop()
+
+    async def cleanup(self):
+        pass
+
+    def link(self, processor: 'FrameProcessor'):
+        self._next = processor
+        processor._prev = self
+        logger.debug(f"Linking {self} -> {self._next}")
+
+    def get_event_loop(self) -> AbstractEventLoop:
+        return self._loop
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        pass
+
+    async def push_frame(self, frame: Frame, direction: FrameDirection = FrameDirection.DOWNSTREAM):
+        if direction == FrameDirection.DOWNSTREAM and self._next:
+            logger.trace(f"Pushing {frame} from {self} to {self._next}")
+            await self._next.process_frame(frame, direction)
+        elif direction == FrameDirection.UPSTREAM and self._prev:
+            logger.trace(f"Pushing {frame} upstream from {self} to {self._prev}")
+            await self._prev.process_frame(frame, direction)
+
+    def __str__(self):
+        return self.name

src/pipecat/processors/logger.py

@@ -0,0 +1,22 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.frames.frames import Frame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class FrameLogger(FrameProcessor):
+    def __init__(self, prefix="Frame"):
+        super().__init__()
+        self._prefix = prefix
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        match direction:
+            case FrameDirection.UPSTREAM:
+                print(f"< {self._prefix}: {frame}")
+            case FrameDirection.DOWNSTREAM:
+                print(f"> {self._prefix}: {frame}")
+        await self.push_frame(frame, direction)

src/pipecat/processors/text_transformer.py

@@ -0,0 +1,36 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from typing import Coroutine
+
+from pipecat.frames.frames import Frame, TextFrame
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class StatelessTextTransformer(FrameProcessor):
+    """This processor calls the given function on any text in a text frame.
+
+    >>> async def print_frames(aggregator, frame):
+    ...     async for frame in aggregator.process_frame(frame):
+    ...         print(frame.text)
+
+    >>> aggregator = StatelessTextTransformer(lambda x: x.upper())
+    >>> asyncio.run(print_frames(aggregator, TextFrame("Hello")))
+    HELLO
+    """
+
+    def __init__(self, transform_fn):
+        super().__init__()
+        self._transform_fn = transform_fn
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            result = self._transform_fn(frame.text)
+            if isinstance(result, Coroutine):
+                result = await result
+            await self.push_frame(result)
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/processors/utils/audio.py

@@ -0,0 +1,25 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from typing import List
+
+from pipecat.frames.frames import AudioRawFrame
+
+
+def maybe_split_audio_frame(frame: AudioRawFrame, largest_write_size: int) -> List[AudioRawFrame]:
+    """Subdivide large audio frames to enable interruption."""
+    frames: List[AudioRawFrame] = []
+    if len(frame.audio) > largest_write_size:
+        for i in range(0, len(frame.audio), largest_write_size):
+            chunk = frame.audio[i: i + largest_write_size]
+            frames.append(
+                AudioRawFrame(
+                    audio=chunk,
+                    sample_rate=frame.sample_rate,
+                    num_channels=frame.num_channels))
+    else:
+        frames.append(frame)
+    return frames

src/pipecat/serializers/abstract_frame_serializer.py

@@ -1,6 +1,6 @@
 from abc import abstractmethod
 
-from dailyai.pipeline.frames import Frame
+from pipecat.pipeline.frames import Frame
 
 
 class FrameSerializer:

src/pipecat/serializers/protobuf_serializer.py

@@ -1,8 +1,8 @@
 import dataclasses
 from typing import Text
-from dailyai.pipeline.frames import AudioFrame, Frame, TextFrame, TranscriptionFrame
-import dailyai.pipeline.protobufs.frames_pb2 as frame_protos
-from dailyai.serializers.abstract_frame_serializer import FrameSerializer
+from pipecat.pipeline.frames import AudioFrame, Frame, TextFrame, TranscriptionFrame
+import pipecat.pipeline.protobufs.frames_pb2 as frame_protos
+from pipecat.serializers.abstract_frame_serializer import FrameSerializer
 
 
 class ProtobufFrameSerializer(FrameSerializer):

src/pipecat/services/ai_services.py

@@ -0,0 +1,171 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import array
+import io
+import math
+import wave
+
+from abc import abstractmethod
+from typing import BinaryIO
+
+from pipecat.frames.frames import (
+    AudioRawFrame,
+    EndFrame,
+    Frame,
+    TextFrame,
+    VisionImageRawFrame,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+
+class AIService(FrameProcessor):
+    def __init__(self):
+        super().__init__()
+
+
+class LLMService(AIService):
+    """This class is a no-op but serves as a base class for LLM services."""
+
+    def __init__(self):
+        super().__init__()
+
+
+class TTSService(AIService):
+    def __init__(self, aggregate_sentences: bool = True):
+        super().__init__()
+        self._aggregate_sentences: bool = aggregate_sentences
+        self._current_sentence: str = ""
+
+    # Converts the text to audio.
+    @abstractmethod
+    async def run_tts(self, text: str):
+        pass
+
+    async def say(self, text: str):
+        await self.process_frame(TextFrame(text=text), FrameDirection.DOWNSTREAM)
+
+    async def _process_text_frame(self, frame: TextFrame):
+        text: str | None = None
+        if not self._aggregate_sentences:
+            text = frame.text
+        else:
+            self._current_sentence += frame.text
+            if self._current_sentence.strip().endswith((".", "?", "!")):
+                text = self._current_sentence
+                self._current_sentence = ""
+
+        if text:
+            await self.run_tts(text)
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            await self._process_text_frame(frame)
+        elif isinstance(frame, EndFrame):
+            if self._current_sentence:
+                await self.run_tts(self._current_sentence)
+            await self.push_frame(frame)
+        else:
+            await self.push_frame(frame, direction)
+
+
+class STTService(AIService):
+    """STTService is a base class for speech-to-text services."""
+
+    def __init__(self,
+                 min_rms: int = 400,
+                 max_silence_frames: int = 3,
+                 sample_rate: int = 16000,
+                 num_channels: int = 1):
+        super().__init__()
+        self._min_rms = min_rms
+        self._max_silence_frames = max_silence_frames
+        self._sample_rate = sample_rate
+        self._num_channels = num_channels
+        self._current_silence_frames = 0
+        (self._content, self._wave) = self._new_wave()
+
+    @abstractmethod
+    async def run_stt(self, audio: BinaryIO):
+        """Returns transcript as a string"""
+        pass
+
+    def _new_wave(self):
+        content = io.BufferedRandom(io.BytesIO())
+        ww = wave.open(content, "wb")
+        ww.setsampwidth(2)
+        ww.setnchannels(self._num_channels)
+        ww.setframerate(self._sample_rate)
+        return (content, ww)
+
+    def _get_volume(self, audio: bytes) -> float:
+        # https://docs.python.org/3/library/array.html
+        audio_array = array.array('h', audio)
+        squares = [sample**2 for sample in audio_array]
+        mean = sum(squares) / len(audio_array)
+        rms = math.sqrt(mean)
+        return rms
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        """Processes a frame of audio data, either buffering or transcribing it."""
+        if not isinstance(frame, AudioRawFrame):
+            await self.push_frame(frame, direction)
+            return
+
+        audio = frame.audio
+
+        # Try to filter out empty background noise
+        # (Very rudimentary approach, can be improved)
+        rms = self._get_volume(audio)
+        if rms >= self._min_rms:
+            # If volume is high enough, write new data to wave file
+            self._wave.writeframes(audio)
+
+        # If buffer is not empty and we detect a 3-frame pause in speech,
+        # transcribe the audio gathered so far.
+        if self._content.tell() > 0 and self._current_silence_frames > self._max_silence_frames:
+            self._current_silence_frames = 0
+            self._wave.close()
+            self._content.seek(0)
+            await self.run_stt(self._content)
+            (self._content, self._wave) = self._new_wave()
+        # If we get this far, this is a frame of silence
+        self._current_silence_frames += 1
+
+
+class ImageGenService(AIService):
+
+    def __init__(self):
+        super().__init__()
+
+    # Renders the image. Returns an Image object.
+    @abstractmethod
+    async def run_image_gen(self, prompt: str):
+        pass
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, TextFrame):
+            await self.run_image_gen(frame.text)
+        else:
+            await self.push_frame(frame, direction)
+
+
+class VisionService(AIService):
+    """VisionService is a base class for vision services."""
+
+    def __init__(self):
+        super().__init__()
+        self._describe_text = None
+
+    @abstractmethod
+    async def run_vision(self, frame: VisionImageRawFrame):
+        pass
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, VisionImageRawFrame):
+            await self.run_vision(frame)
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/services/anthropic.py

@@ -0,0 +1,51 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.frames.frames import Frame, LLMMessagesFrame, TextFrame
+from pipecat.processors.frame_processor import FrameDirection
+from pipecat.services.ai_services import LLMService
+
+from loguru import logger
+
+try:
+    from anthropic import AsyncAnthropic
+except ModuleNotFoundError as e:
+    logger.error(f"Exception: {e}")
+    logger.error(
+        "In order to use Anthropic, you need to `pip install pipecat-ai[anthropic]`. Also, set `ANTHROPIC_API_KEY` environment variable.")
+    raise Exception(f"Missing module: {e}")
+
+
+class AnthropicLLMService(LLMService):
+
+    def __init__(
+            self,
+            api_key,
+            model="claude-3-opus-20240229",
+            max_tokens=1024):
+        super().__init__()
+        self.client = AsyncAnthropic(api_key=api_key)
+        self.model = model
+        self.max_tokens = max_tokens
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        if isinstance(frame, LLMMessagesFrame):
+            stream = await self.client.messages.create(
+                max_tokens=self.max_tokens,
+                messages=[
+                    {
+                        "role": "user",
+                        "content": "Hello, Claude",
+                    }
+                ],
+                model=self.model,
+                stream=True,
+            )
+            async for event in stream:
+                if event.type == "content_block_delta":
+                    await self.push_frame(TextFrame(event.delta.text))
+        else:
+            await self.push_frame(frame, direction)

src/pipecat/services/azure.py

@@ -5,9 +5,11 @@ from openai import AsyncAzureOpenAI
 
 from collections.abc import AsyncGenerator
 
-from dailyai.services.ai_services import TTSService, ImageGenService
+from pipecat.services.ai_services import TTSService, ImageGenService
 from PIL import Image
 
+from loguru import logger
+
 # See .env.example for Azure configuration needed
 try:
     from azure.cognitiveservices.speech import (
@@ -17,12 +19,12 @@ try:
         CancellationReason,
     )
 except ModuleNotFoundError as e:
-    print(f"Exception: {e}")
-    print(
-        "In order to use Azure TTS, you need to `pip install dailyai[azure]`. Also, set `SPEECH_KEY` and `SPEECH_REGION` environment variables.")
+    logger.error(f"Exception: {e}")
+    logger.error(
+        "In order to use Azure TTS, you need to `pip install pipecat-ai[azure]`. Also, set `AZURE_SPEECH_API_KEY` and `AZURE_SPEECH_REGION` environment variables.")
     raise Exception(f"Missing module: {e}")
 
-from dailyai.services.openai_api_llm_service import BaseOpenAILLMService
+from pipecat.services.openai_api_llm_service import BaseOpenAILLMService
 
 
 class AzureTTSService(TTSService):
@@ -97,23 +99,24 @@ class AzureImageGenServiceREST(ImageGenService):
         endpoint,
         model,
     ):
-        super().__init__(image_size=image_size)
+        super().__init__()
 
         self._api_key = api_key
         self._azure_endpoint = endpoint
         self._api_version = api_version
         self._model = model
         self._aiohttp_session = aiohttp_session
+        self._image_size = image_size
 
-    async def run_image_gen(self, sentence) -> tuple[str, bytes]:
+    async def run_image_gen(self, prompt: str) -> tuple[str, bytes, tuple[int, int]]:
         url = f"{self._azure_endpoint}openai/images/generations:submit?api-version={self._api_version}"
         headers = {
             "api-key": self._api_key,
             "Content-Type": "application/json"}
         body = {
             # Enter your prompt text here
-            "prompt": sentence,
-            "size": self.image_size,
+            "prompt": prompt,
+            "size": self._image_size,
             "n": 1,
         }
         async with self._aiohttp_session.post(
@@ -146,4 +149,4 @@ class AzureImageGenServiceREST(ImageGenService):
             async with self._aiohttp_session.get(image_url) as response:
                 image_stream = io.BytesIO(await response.content.read())
                 image = Image.open(image_stream)
-                return (image_url, image.tobytes())
+                return (image_url, image.tobytes(), image.size)

src/pipecat/services/deepgram.py

@@ -1,8 +1,17 @@
-from collections.abc import AsyncGenerator
-from dailyai.services.ai_services import TTSService
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.frames.frames import AudioRawFrame
+from pipecat.services.ai_services import TTSService
+
+from loguru import logger
 
 
 class DeepgramTTSService(TTSService):
+
     def __init__(
             self,
             *,
@@ -15,15 +24,13 @@ class DeepgramTTSService(TTSService):
         self._api_key = api_key
         self._aiohttp_session = aiohttp_session
 
-    def get_mic_sample_rate(self):
-        return 24000
-
-    async def run_tts(self, sentence) -> AsyncGenerator[bytes, None]:
-        self.logger.info(f"Running deepgram tts for {sentence}")
+    async def run_tts(self, text: str):
+        logger.info(f"Running Deepgram TTS for {text}")
         base_url = "https://api.beta.deepgram.com/v1/speak"
         request_url = f"{base_url}?model={self._voice}&encoding=linear16&container=none&sample_rate=16000"
         headers = {"authorization": f"token {self._api_key}"}
-        body = {"text": sentence}
+        body = {"text": text}
         async with self._aiohttp_session.post(request_url, headers=headers, json=body) as r:
             async for data in r.content:
-                yield data
+                frame = AudioRawFrame(audio=data, sample_rate=16000, num_channels=1)
+                await self.push_frame(frame)

src/pipecat/services/elevenlabs.py

@@ -0,0 +1,58 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import aiohttp
+
+from pipecat.frames.frames import AudioRawFrame, TTSStartedFrame, TTSStoppedFrame
+from pipecat.services.ai_services import TTSService
+
+from loguru import logger
+
+
+class ElevenLabsTTSService(TTSService):
+
+    def __init__(
+        self,
+        *,
+        aiohttp_session: aiohttp.ClientSession,
+        api_key: str,
+        voice_id: str,
+        model: str = "eleven_turbo_v2",
+    ):
+        super().__init__()
+
+        self._api_key = api_key
+        self._voice_id = voice_id
+        self._aiohttp_session = aiohttp_session
+        self._model = model
+
+    async def run_tts(self, text: str):
+        logger.debug(f"Transcribing text: {text}")
+
+        url = f"https://api.elevenlabs.io/v1/text-to-speech/{self._voice_id}/stream"
+
+        payload = {"text": text, "model_id": self._model}
+
+        querystring = {
+            "output_format": "pcm_16000",
+            "optimize_streaming_latency": 2}
+
+        headers = {
+            "xi-api-key": self._api_key,
+            "Content-Type": "application/json",
+        }
+
+        async with self._aiohttp_session.post(url, json=payload, headers=headers, params=querystring) as r:
+            if r.status != 200:
+                logger.error(f"Audio fetch status code: {r.status}, error: {r.text}")
+                return
+
+            await self.push_frame(TTSStartedFrame())
+            async for chunk in r.content:
+                if len(chunk) > 0:
+                    frame = AudioRawFrame(chunk, 16000, 1)
+                    await self.push_frame(frame)
+            await self.push_frame(TTSStoppedFrame())

src/pipecat/services/fal.py

@@ -0,0 +1,83 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import aiohttp
+import io
+import os
+
+from PIL import Image
+from numpy import result_type
+from pydantic import BaseModel
+from typing import Optional, Union, Dict
+
+from pipecat.frames.frames import URLImageRawFrame
+from pipecat.services.ai_services import ImageGenService
+
+from loguru import logger
+
+try:
+    import fal_client
+except ModuleNotFoundError as e:
+    logger.error(f"Exception: {e}")
+    logger.error(
+        "In order to use Fal, you need to `pip install pipecat-ai[fal]`. Also, set `FAL_KEY` environment variable.")
+    raise Exception(f"Missing module: {e}")
+
+
+class FalImageGenService(ImageGenService):
+    class InputParams(BaseModel):
+        seed: Optional[int] = None
+        num_inference_steps: int = 8
+        num_images: int = 1
+        image_size: Union[str, Dict[str, int]] = "square_hd"
+        expand_prompt: bool = False
+        enable_safety_checker: bool = True
+        format: str = "png"
+
+    def __init__(
+        self,
+        *,
+        aiohttp_session: aiohttp.ClientSession,
+        params: InputParams,
+        model: str = "fal-ai/fast-sdxl",
+        key: str | None = None,
+    ):
+        super().__init__()
+        self._model = model
+        self._params = params
+        self._aiohttp_session = aiohttp_session
+        if key:
+            os.environ["FAL_KEY"] = key
+
+    async def run_image_gen(self, prompt: str):
+        logger.debug(f"Generating image from prompt: {prompt}")
+
+        response = await fal_client.run_async(
+            self._model,
+            arguments={"prompt": prompt, **self._params.model_dump()}
+        )
+
+        image_url = response["images"][0]["url"] if response else None
+
+        if not image_url:
+            logger.error("Image generation failed")
+            return
+
+        logger.debug(f"Image generated at: {image_url}")
+
+        # Load the image from the url
+        logger.debug(f"Downloading image {image_url} ...")
+        async with self._aiohttp_session.get(image_url) as response:
+            logger.debug(f"Downloaded image {image_url}")
+            image_stream = io.BytesIO(await response.content.read())
+            image = Image.open(image_stream)
+
+            frame = URLImageRawFrame(
+                url=image_url,
+                image=image.tobytes(),
+                size=image.size,
+                format=image.format)
+            await self.push_frame(frame)

src/pipecat/services/fireworks.py

@@ -0,0 +1,24 @@
+#
+# Copyright (c) 2024, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+from pipecat.services.openai import BaseOpenAILLMService
+
+from loguru import logger
+
+try:
+    from openai import AsyncOpenAI
+except ModuleNotFoundError as e:
+    logger.error(f"Exception: {e}")
+    logger.error(
+        "In order to use Fireworks, you need to `pip install pipecat-ai[fireworks]`. Also, set the `FIREWORKS_API_KEY` environment variable.")
+    raise Exception(f"Missing module: {e}")
+
+
+class FireworksLLMService(BaseOpenAILLMService):
+    def __init__(self,
+                 model="accounts/fireworks/models/firefunction-v1",
+                 base_url="https://api.fireworks.ai/inference/v1"):
+        super().__init__(model, base_url)