Reorganize examples into topic-based subfolders

Move 304 examples from a flat numbered directory into 14 descriptive
subfolders: getting-started, services (speech + function-calling),
transcription, vision, realtime, persistent-context,
context-summarization, update-settings (stt/tts/llm), turn-management,
thinking-and-mcp, transports, video-avatar, video-processing, and
features.

Strip numbered prefixes from filenames (e.g. 07c-interruptible-deepgram.py
becomes services/speech/deepgram.py) since the folder context makes them
redundant. Keep numbered prefixes only in getting-started/ where ordering
matters.

Update eval script paths and README to match the new structure.

examples/features/audio-recording.py

@@ -0,0 +1,206 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Audio Recording Example with Pipecat.
+
+This example demonstrates how to record audio from a conversation between a user and an AI assistant,
+saving both merged and individual audio tracks. It showcases the AudioBufferProcessor's capabilities
+to handle both combined and separate audio streams.
+
+The example:
+    1. Sets up a basic conversation with an AI assistant
+    2. Records the entire conversation
+    3. Saves three separate WAV files:
+        - A merged recording of both participants
+        - Individual recording of user audio
+        - Individual recording of assistant audio
+
+Requirements:
+    - OpenAI API key (for GPT-4)
+    - Cartesia API key (for text-to-speech)
+    - Daily API key (for video/audio transport)
+
+    Environment variables (.env file):
+        OPENAI_API_KEY=your_openai_key
+        CARTESIA_API_KEY=your_cartesia_key
+        DAILY_API_KEY=your_daily_key
+        DEEPGRAM_API_KEY=your_deepgram_key
+
+The recordings will be saved in a 'recordings' directory with timestamps:
+    recordings/
+        merged_20240315_123456.wav  (Combined audio)
+        user_20240315_123456.wav    (User audio only)
+        bot_20240315_123456.wav     (Bot audio only)
+
+Note:
+    This example requires the AudioBufferProcessor with track-specific audio support,
+    which provides both 'on_audio_data' and 'on_track_audio_data' events for
+    handling merged and separate audio tracks respectively.
+"""
+
+import datetime
+import io
+import os
+import wave
+
+import aiofiles
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.audio.audio_buffer_processor import AudioBufferProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+async def save_audio_file(audio: bytes, filename: str, sample_rate: int, num_channels: int):
+    """Save audio data to a WAV file."""
+    if len(audio) > 0:
+        with io.BytesIO() as buffer:
+            with wave.open(buffer, "wb") as wf:
+                wf.setsampwidth(2)
+                wf.setnchannels(num_channels)
+                wf.setframerate(sample_rate)
+                wf.writeframes(audio)
+            async with aiofiles.open(filename, "wb") as file:
+                await file.write(buffer.getvalue())
+        logger.info(f"Audio saved to {filename}")
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"), audio_passthrough=True)
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    # Create audio buffer processor
+    audiobuffer = AudioBufferProcessor()
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            audiobuffer,  # Add audio buffer to pipeline
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Start recording audio
+        await audiobuffer.start_recording()
+        # Start conversation - empty prompt to let LLM follow system instructions
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    # Handler for merged audio
+    @audiobuffer.event_handler("on_audio_data")
+    async def on_audio_data(buffer, audio, sample_rate, num_channels):
+        timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+        filename = f"recordings/merged_{timestamp}.wav"
+        os.makedirs("recordings", exist_ok=True)
+        await save_audio_file(audio, filename, sample_rate, num_channels)
+
+    # Handler for separate tracks
+    @audiobuffer.event_handler("on_track_audio_data")
+    async def on_track_audio_data(buffer, user_audio, bot_audio, sample_rate, num_channels):
+        timestamp = datetime.datetime.now().strftime("%Y%m%d_%H%M%S")
+        os.makedirs("recordings", exist_ok=True)
+
+        # Save user audio
+        user_filename = f"recordings/user_{timestamp}.wav"
+        await save_audio_file(user_audio, user_filename, sample_rate, 1)
+
+        # Save bot audio
+        bot_filename = f"recordings/bot_{timestamp}.wav"
+        await save_audio_file(bot_audio, bot_filename, sample_rate, 1)
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/before-and-after-events.py

@@ -0,0 +1,148 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+from dataclasses import dataclass
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import DataFrame, LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+@dataclass
+class CustomBeforeProcessFrame(DataFrame):
+    pass
+
+
+@dataclass
+class CustomAfterPushFrame(DataFrame):
+    pass
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.output().event_handler("on_before_process_frame")
+    async def transport_on_before_process_frame(transport, frame):
+        if isinstance(frame, CustomBeforeProcessFrame):
+            logger.info("*** CUSTOM FRAME RECEIVED. Bot is about to talk, let's do something!")
+
+    @transport.output().event_handler("on_after_push_frame")
+    async def transport_on_after_push_frame(transport, frame):
+        if isinstance(frame, CustomAfterPushFrame):
+            logger.info("*** CUSTOM FRAME RECEIVED. Bot has stopped talking, let's do something!")
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        # Custom frames are pushed in order so they can be used for synchronization purposes.
+        await task.queue_frames([CustomBeforeProcessFrame(), LLMRunFrame(), CustomAfterPushFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/bot-background-sound.py

@@ -0,0 +1,155 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.mixers.soundfile_mixer import SoundfileMixer
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame, MixerEnableFrame, MixerUpdateSettingsFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+OFFICE_SOUND_FILE = os.path.join(
+    os.path.dirname(__file__), "assets", "office-ambience-24000-mono.mp3"
+)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        audio_out_mixer=SoundfileMixer(
+            sound_files={"office": OFFICE_SOUND_FILE},
+            default_sound="office",
+            volume=2.0,
+        ),
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        audio_out_mixer=SoundfileMixer(
+            sound_files={"office": OFFICE_SOUND_FILE},
+            default_sound="office",
+            volume=2.0,
+        ),
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        audio_out_mixer=SoundfileMixer(
+            sound_files={"office": OFFICE_SOUND_FILE},
+            default_sound="office",
+            volume=2.0,
+        ),
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, participant):
+        # Show how to use mixer control frames.
+        logger.info(f"Listening for background sound for a bit...")
+        await asyncio.sleep(5.0)
+        logger.info(f"Reducing volume...")
+        await task.queue_frame(MixerUpdateSettingsFrame({"volume": 0.5}))
+        await asyncio.sleep(5.0)
+        logger.info(f"Disabling background sound for a bit...")
+        await task.queue_frame(MixerEnableFrame(False))
+        await asyncio.sleep(5.0)
+        logger.info(f"Re-enabling background sound and starting bot...")
+        await task.queue_frame(MixerEnableFrame(True))
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/concurrent-llm-evaluation.py

@@ -0,0 +1,173 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.audio.vad_processor import VADProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.groq.llm import GroqLLMService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_turn_processor import UserTurnProcessor
+from pipecat.turns.user_turn_strategies import ExternalUserTurnStrategies
+
+load_dotenv(override=True)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    openai_llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    groq_llm = GroqLLMService(
+        api_key=os.getenv("GROQ_API_KEY"),
+        settings=GroqLLMService.Settings(
+            system_instruction="You are a very helpful assistant. Your goal is to demonstrate your capabilities in detail in a creative and helpful way.",
+        ),
+    )
+
+    openai_context = LLMContext()
+    groq_context = LLMContext()
+
+    # We use an external VADProcessor because the UserTurnProcessor is shared
+    # across multiple parallel aggregators. The VADProcessor emits
+    # VADUserStartedSpeakingFrame and VADUserStoppedSpeakingFrame which the
+    # UserTurnProcessor needs to manage turn lifecycle.
+    vad_processor = VADProcessor(vad_analyzer=SileroVADAnalyzer())
+
+    # We use this external user turn processor. This processor will push
+    # UserStartedSpeakingFrame and UserStoppedSpeakingFrame as well as
+    # interruptions. This can be used in advanced cases when there are multiple
+    # aggregators in the pipeline.
+    user_turn_processor = UserTurnProcessor()
+
+    # We use external user turn strategies for both aggregators since the turn
+    # management is done by the common UserTurnProcessor.
+    openai_context_aggregator = LLMContextAggregatorPair(
+        openai_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+    groq_context_aggregator = LLMContextAggregatorPair(
+        groq_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            vad_processor,
+            user_turn_processor,
+            ParallelPipeline(
+                [
+                    openai_context_aggregator.user(),  # User responses
+                    openai_llm,  # LLM
+                    tts,  # TTS (bot will speak the chosen language)
+                    transport.output(),  # Transport bot output
+                    openai_context_aggregator.assistant(),  # Assistant spoken responses
+                ],
+                [
+                    groq_context_aggregator.user(),  # User responses
+                    groq_llm,  # LLM
+                    groq_context_aggregator.assistant(),  # Assistant responses
+                ],
+            ),
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        openai_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        groq_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/concurrent-llm-rtvi-ignored-sources.py

@@ -0,0 +1,184 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""RTVIObserver ignored sources example.
+
+This example shows how to suppress RTVI messages from a specific pipeline
+processor so that secondary branches don't leak events to the client.
+
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.audio.vad_processor import VADProcessor
+from pipecat.processors.frameworks.rtvi import RTVIObserverParams
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_turn_processor import UserTurnProcessor
+from pipecat.turns.user_turn_strategies import ExternalUserTurnStrategies
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info("Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    # Main LLM — drives the conversation. Its RTVI events reach the client.
+    main_llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    # Evaluator LLM — silently grades the user's message in the background.
+    # Its RTVI events will be suppressed so the client is unaware of this branch.
+    evaluator_llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        name="EvaluatorLLM",
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a silent quality evaluator. When given a user message, respond with a single JSON object: {'score': <1-5>, 'reason': '<brief reason>'}. Do not respond conversationally.",
+        ),
+    )
+
+    main_context = LLMContext()
+    evaluator_context = LLMContext()
+
+    # We use an external VADProcessor because the UserTurnProcessor is shared
+    # across multiple parallel aggregators. The VADProcessor emits
+    # VADUserStartedSpeakingFrame and VADUserStoppedSpeakingFrame which the
+    # UserTurnProcessor needs to manage turn lifecycle.
+    vad_processor = VADProcessor(vad_analyzer=SileroVADAnalyzer())
+
+    # We use this external user turn processor. This processor will push
+    # UserStartedSpeakingFrame and UserStoppedSpeakingFrame as well as
+    # interruptions. This can be used in advanced cases when there are multiple
+    # aggregators in the pipeline.
+    user_turn_processor = UserTurnProcessor()
+
+    # We use external user turn strategies for both aggregators since the turn
+    # management is done by the common UserTurnProcessor.
+    main_context_aggregator = LLMContextAggregatorPair(
+        main_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+    evaluator_context_aggregator = LLMContextAggregatorPair(
+        evaluator_context,
+        user_params=LLMUserAggregatorParams(user_turn_strategies=ExternalUserTurnStrategies()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            vad_processor,
+            user_turn_processor,
+            ParallelPipeline(
+                # Main branch: speaks to the user.
+                [
+                    main_context_aggregator.user(),
+                    main_llm,
+                    tts,
+                    transport.output(),
+                    main_context_aggregator.assistant(),
+                ],
+                # Evaluator branch: silent background scoring, no audio output.
+                [
+                    evaluator_context_aggregator.user(),
+                    evaluator_llm,
+                    evaluator_context_aggregator.assistant(),
+                ],
+            ),
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        rtvi_observer_params=RTVIObserverParams(ignored_sources=[evaluator_llm]),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info("Client connected")
+        main_context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        evaluator_context.add_message(
+            {"role": "developer", "content": "Ready to evaluate user messages."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info("Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/custom-frame-processor.py

@@ -0,0 +1,168 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import (
+    Frame,
+    LLMRunFrame,
+    MetricsFrame,
+)
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+
+load_dotenv(override=True)
+
+
+def format_metrics(metrics, indent=0):
+    lines = []
+    tab = "\t" * indent
+
+    for metric in metrics:
+        lines.append(tab + type(metric).__name__)
+        for field, value in vars(metric).items():
+            if hasattr(value, "__dict__") and not isinstance(
+                value, (str, int, float, bool, type(None))
+            ):
+                lines.append(f"{tab}\t{field}={type(value).__name__}")
+                for k, v in vars(value).items():
+                    lines.append(f"{tab}\t\t{k}={repr(v)}")
+            else:
+                lines.append(f"{tab}\t{field}={repr(value)}")
+
+    return "\n".join(lines)
+
+
+class MetricsFrameLogger(FrameProcessor):
+    """MetricsFrameLogger formats and logs all MetericsFrames"""
+
+    def __init__(self, **kwargs):
+        super().__init__(**kwargs)
+
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, MetricsFrame):
+            logger.info(f"{frame.name}\n    {format_metrics(frame.data)}")
+            await self.push_frame(frame, direction)
+
+        # ALWAYS push all frames
+        else:
+            # SUPER IMPORTANT: always push every frame!
+            await self.push_frame(frame, direction)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+        video_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    metrics_frame_processor = MetricsFrameLogger()
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+            metrics_frame_processor,  # pretty print metrics frames
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected: {client}")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/gemini-grounding-metadata.py

@@ -0,0 +1,175 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+import sys
+from pathlib import Path
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.observers.base_observer import BaseObserver, FramePushed
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.google.llm import GoogleLLMService, LLMSearchResponseFrame
+from pipecat.services.llm_service import LLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+sys.path.append(str(Path(__file__).parent.parent))
+
+load_dotenv(override=True)
+
+
+# Function handlers for the LLM
+search_tool = {"google_search": {}}
+tools = [search_tool]
+
+system_instruction = """
+You are an expert at providing the most recent news from any place. Your responses will be converted to audio, so avoid using special characters or overly complex formatting.
+
+Always use the google search API to retrieve the latest news. You must also use it to check which day is today.
+
+You can:
+- Use the Google search API to check the current date.
+- Provide the most recent and relevant news from any place by using the google search API.
+- Answer any questions the user may have, ensuring your responses are accurate and concise.
+
+Start each interaction by asking the user about which place they would like to know the information.
+"""
+
+
+class LLMSearchLoggerObserver(BaseObserver):
+    async def on_push_frame(self, data: FramePushed):
+        src = data.source
+        dst = data.destination
+        frame = data.frame
+        timestamp = data.timestamp
+
+        if not isinstance(src, LLMService) and not isinstance(dst, LLMService):
+            return
+
+        time_sec = timestamp / 1_000_000_000
+
+        arrow = "→"
+
+        if isinstance(frame, LLMSearchResponseFrame):
+            logger.debug(f"🧠 {arrow} {dst} LLM SEARCH RESPONSE FRAME: {frame} at {time_sec:.2f}s")
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    # Initialize the Gemini Multimodal Live model
+    llm = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
+        settings=GoogleLLMService.Settings(
+            system_instruction=system_instruction,
+        ),
+        tools=tools,
+    )
+
+    context = LLMContext(
+        [
+            {
+                "role": "developer",
+                "content": "Start by greeting the user warmly, introducing yourself, and mentioning the current day. Be friendly and engaging to set a positive tone for the interaction.",
+            }
+        ],
+    )
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+        observers=[LLMSearchLoggerObserver()],
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Start conversation - empty prompt to let LLM follow system instructions
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/gemini-rag.py

@@ -0,0 +1,274 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+"""CrossFit Games 2025 Rulebook RAG Demo.
+
+This example demonstrates a Model-Assisted Generation (MAG) chatbot using Google's Gemini model.
+This example uses 2 Gemini models:
+- Gemini 2.0 Flash: This is the voice model that is used to generate the response.
+- Gemini 2.0 Flash Lite: This is the model that is used to answer questions about the CrossFit Games 2025 rulebook - information that isn't yet publicly
+indexed by Gemini (or any other LLM).
+
+How it works:
+- The voice model (Gemini 2.0 Flash) is configured to call a function whenever the user asks a question.
+- The function call is a tool call to the MAG model (Gemini 2.0 Flash Lite).
+- The MAG model generates a response based on the question. The MAG model has the entire contents of the CrossFit Games 2025 rulebook in it's context window.
+- The response is returned to the voice model (Gemini 2.0 Flash), which then generates the response to the user.
+
+Why this works:
+- Gemini 2.0 Flash is fast
+- Gemini 2.0 Flash Lite is faster
+- Gemini 2.0 Flash Lite has a large (1 million tokens) context window
+- IMPORTANT: The generated response from Gemini 2.0 Flash Lite is limited to 50 words or less and 64 tokens.
+You can see this in the RAG_PROMPT variable and the generation_config in the query_knowledge_base function.
+Long generations are slower and more expensive, in the world of Voice AI, we don't need long generations.
+
+Example questions to ask and compare to other RAG solutions:
+- What lenses are not allowed?
+- How many people can be on a team?
+- What do winning gyms get?
+- What happens if I skip a workout?
+- Can I switch my team members for the Games?
+- What happens if I start too early?
+
+Notes:
+- The RAG model is Gemini 2.0 Flash Lite.
+- The voice model is Gemini 2.0 Flash.
+- The RAG content is stored in the assets/rag-content.txt file.
+- The model for voice is Gemini 2.0 Flash, but can be easily switched to any other model.
+
+Customization options:
+- update assets/rag-content.txt with your own knowledge base
+- increase/decrease the RAG_MODEL's generation length
+- use a different voice model
+- play with the RAG_PROMPT
+- change the function calling logic
+"""
+
+import json
+import os
+import time
+
+from dotenv import load_dotenv
+from google import genai
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.google.llm import GoogleLLMService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# Initialize the client globally
+client = genai.Client(api_key=os.environ["GOOGLE_API_KEY"])
+
+
+def get_rag_content():
+    """Get the RAG content from the file."""
+    script_dir = os.path.dirname(os.path.abspath(__file__))
+    rag_content_path = os.path.join(script_dir, "assets", "rag-content.txt")
+    with open(rag_content_path, "r") as f:
+        return f.read()
+
+
+RAG_MODEL = "gemini-2.0-flash-lite-preview-02-05"
+VOICE_MODEL = "gemini-2.0-flash"
+RAG_CONTENT = get_rag_content()
+RAG_PROMPT = f"""
+You are a helpful assistant designed to answer user questions based solely on the provided knowledge base.
+
+**Instructions:**
+
+1.  **Knowledge Base Only:** Answer questions *exclusively* using the information in the "Knowledge Base" section below. Do not use any outside information.
+2.  **Conversation History:** Use the "Conversation History" (ordered oldest to newest) to understand the context of the current question.
+3.  **Concise Response:**  Respond in 50 words or fewer.  The response will be spoken, so avoid symbols, abbreviations, or complex formatting. Use plain, natural language.
+4.  **Unknown Answer:** If the answer is not found within the "Knowledge Base," respond with "I don't know." Do not guess or make up an answer.
+5. Do not introduce your response. Just provide the answer.
+6. You must follow all instructions.
+
+**Input Format:**
+
+Each request will include:
+
+*   **Conversation History:**  (A list of previous user and assistant messages, if any)
+
+**Knowledge Base:**
+Here is the knowledge base you have access to:
+{RAG_CONTENT}
+"""
+
+
+async def query_knowledge_base(params: FunctionCallParams):
+    """Query the knowledge base for the answer to the question."""
+    logger.info(f"Querying knowledge base for question: {params.arguments['question']}")
+
+    # for our case, the first two messages are the instructions and the user message
+    # so we remove them.
+    conversation_turns = params.context.get_messages()[2:]
+
+    def _is_tool_call(turn):
+        if turn.get("role", None) == "tool":
+            return True
+        if turn.get("tool_calls", None):
+            return True
+        return False
+
+    # filter out tool calls
+    messages = [turn for turn in conversation_turns if not _is_tool_call(turn)]
+    # use the last 3 turns as the conversation history/context
+    messages = messages[-3:]
+    messages_json = json.dumps(messages, ensure_ascii=False, indent=2)
+
+    logger.info(f"Conversation turns: {messages_json}")
+
+    start = time.perf_counter()
+    full_prompt = f"System: {RAG_PROMPT}\n\nConversation History: {messages_json}"
+
+    response = await client.aio.models.generate_content(
+        model=RAG_MODEL,
+        contents=[full_prompt],
+        config={
+            "temperature": 0.1,
+            "max_output_tokens": 64,
+        },
+    )
+    end = time.perf_counter()
+    logger.info(f"Time taken: {end - start:.2f} seconds")
+    logger.info(response.text)
+    await params.result_callback(response.text)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="f9836c6e-a0bd-460e-9d3c-f7299fa60f94",  # Southern Lady
+        ),
+    )
+
+    system_prompt = """\
+You are a helpful assistant who converses with a user and answers questions.
+
+You have access to the tool, query_knowledge_base, that allows you to query the knowledge base for the answer to the user's question.
+
+Your response will be turned into speech so use only simple words and punctuation.
+"""
+
+    llm = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
+        settings=GoogleLLMService.Settings(
+            model=VOICE_MODEL,
+            system_instruction=system_prompt,
+        ),
+    )
+    llm.register_function("query_knowledge_base", query_knowledge_base)
+
+    query_function = FunctionSchema(
+        name="query_knowledge_base",
+        description="Query the knowledge base for the answer to the question.",
+        properties={
+            "question": {
+                "type": "string",
+                "description": "The question to query the knowledge base with.",
+            },
+        },
+        required=["question"],
+    )
+    tools = ToolsSchema(standard_tools=[query_function])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Start conversation - empty prompt to let LLM follow system instructions
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/gpu-container-local-bot.py

@@ -0,0 +1,165 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.deepgram.tts import DeepgramTTSService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import (
+    DailyOutputTransportMessageFrame,
+    DailyOutputTransportMessageUrgentFrame,
+    DailyParams,
+)
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = DeepgramTTSService(
+        api_key=os.getenv("DEEPGRAM_API_KEY"),
+        settings=DeepgramTTSService.Settings(
+            voice="aura-asteria-en",
+        ),
+        base_url="http://0.0.0.0:8080",
+    )
+
+    llm = OpenAILLMService(
+        # To use OpenAI
+        # api_key=os.getenv("OPENAI_API_KEY"),
+        # Or, to use a local vLLM (or similar) api server
+        settings=OpenAILLMService.Settings(
+            model="meta-llama/Meta-Llama-3-8B-Instruct",
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+        base_url="http://0.0.0.0:8000/v1",
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            user_aggregator,
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    # When the first participant joins, the bot should introduce itself.
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    # Handle "latency-ping" messages. The client will send app messages that look like
+    # this:
+    #   { "latency-ping": { ts: <client-side timestamp> }}
+    #
+    # We want to send an immediate pong back to the client from this handler function.
+    # Also, we will push a frame into the top of the pipeline and send it after the
+    #
+    @transport.event_handler("on_app_message")
+    async def on_app_message(transport, message, sender):
+        try:
+            if "latency-ping" in message:
+                logger.debug(f"Received latency ping app message: {message}")
+                ts = message["latency-ping"]["ts"]
+                # Send immediately
+                await task.queue_frame(
+                    DailyOutputTransportMessageUrgentFrame(
+                        message={"latency-pong-msg-handler": {"ts": ts}}, participant_id=sender
+                    )
+                )
+                # And push to the pipeline for the Daily transport.output to send
+                await task.queue_frame(
+                    DailyOutputTransportMessageFrame(
+                        message={"latency-pong-pipeline-delivery": {"ts": ts}},
+                        participant_id=sender,
+                    )
+                )
+        except Exception as e:
+            logger.debug(f"message handling error: {e} - {message}")
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/heartbeats.py

@@ -0,0 +1,43 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import sys
+
+from loguru import logger
+
+from pipecat.frames.frames import Frame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+
+logger.remove(0)
+logger.add(sys.stderr, level="DEBUG")
+
+
+class NullProcessor(FrameProcessor):
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        await super().process_frame(frame, direction)
+
+
+async def main():
+    """This test shows heartbeat monitoring.
+
+    A warning is dispalyed when heartbeats are not received within the
+    default (5 seconds) timeout.
+    """
+    pipeline = Pipeline([NullProcessor()])
+
+    task = PipelineTask(pipeline, params=PipelineParams(enable_heartbeats=True))
+
+    runner = PipelineRunner()
+
+    await runner.run(task)
+
+
+if __name__ == "__main__":
+    asyncio.run(main())

examples/features/live-translation.py

@@ -0,0 +1,142 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import TTSSpeakFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_start import TranscriptionUserTurnStartStrategy
+from pipecat.turns.user_turn_strategies import UserTurnStrategies
+
+load_dotenv(override=True)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="d4db5fb9-f44b-4bd1-85fa-192e0f0d75f9",  # Spanish-speaking Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a live translation assistant. Your sole purpose is to translate English text into Spanish. When you receive English text from the user, immediately translate it into natural, fluent Spanish. Do not add explanations, commentary, or extra information—only provide the Spanish translation of the text you receive.",
+        ),
+    )
+
+    context = LLMContext()
+
+    # We use the TranscriptionUserTurnStartStrategy to start a new user turn
+    # every time a transcription is received. We disable interruptions, so the
+    # user can continue speaking while the bot is transcribing, without
+    # interrupting the bot.
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            user_turn_strategies=UserTurnStrategies(
+                start=[TranscriptionUserTurnStartStrategy(enable_interruptions=False)],
+            ),
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS (bot will speak the chosen language)
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        await task.queue_frames(
+            [
+                TTSSpeakFrame(
+                    text="Hello, welcome to live translation. Everything you say will be automatically translated to Spanish. Let's begin!",
+                    append_to_context=True,
+                ),
+            ]
+        )
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/mem0.py

@@ -0,0 +1,270 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Mem0 Personalized Voice Agent Example with Pipecat.
+
+This example demonstrates how to create a conversational AI assistant with memory capabilities
+using Mem0 integration. It shows how to build an agent that remembers previous interactions
+and personalizes responses based on conversation history.
+
+The example:
+    1. Sets up a video/audio conversation between a user and an AI assistant
+    2. Uses Mem0 to store and retrieve memories from conversations
+    3. Creates personalized greetings based on previous interactions
+    4. Handles multi-modal interaction through audio
+    5. Demonstrates two approaches for memory management:
+       - Using Mem0 API (cloud-based memory storage)
+       - Using local configuration with custom LLM (self-hosted memory)
+
+Requirements:
+    - OpenAI API key
+    - ElevenLabs API key (for text-to-speech)
+    - Daily API key (for video/audio transport)
+    - Mem0 API key (for cloud-based memory storage)
+    - [Optional] Anthropic API key (if using Claude with local config)
+
+    Environment variables (set in .env or in your terminal using `export`):
+        DAILY_ROOM_URL=daily_room_url
+        DAILY_API_KEY=daily_api_key
+        OPENAI_API_KEY=openai_api_key
+        ELEVENLABS_API_KEY=elevenlabs_api_key
+        MEM0_API_KEY=mem0_api_key
+        ANTHROPIC_API_KEY=anthropic_api_key (if using Claude with local config)
+
+The bot runs as part of a pipeline that processes audio frames and manages the conversation flow.
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.elevenlabs.tts import ElevenLabsTTSService
+from pipecat.services.mem0.memory import Mem0MemoryService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+async def get_initial_greeting(memory_service: Mem0MemoryService) -> str:
+    """Fetch all memories for the user and create a personalized greeting.
+
+    Args:
+        memory_service: The Mem0 memory service instance.
+
+    Returns:
+        A personalized greeting based on user memories.
+    """
+    try:
+        results = await memory_service.get_memories()
+        if not results:
+            logger.debug("No memories found for this user.")
+            return "Hello! It's nice to meet you. How can I help you today?"
+
+        # Create a personalized greeting based on memories
+        greeting = "Hello! It's great to see you again. "
+        greeting += "Based on our previous conversations, I remember: "
+        for i, memory in enumerate(results[:3], 1):
+            memory_content = memory.get("memory", "")
+            # Keep memory references brief
+            if len(memory_content) > 100:
+                memory_content = memory_content[:97] + "..."
+            greeting += f"{memory_content} "
+
+        greeting += "How can I help you today?"
+
+        logger.debug(f"Created personalized greeting from {len(results)} memories")
+        return greeting
+
+    except Exception as e:
+        logger.error(f"Error retrieving initial memories from Mem0: {e}")
+        return "Hello! How can I help you today?"
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    """Main bot execution function.
+
+    Sets up and runs the bot pipeline including:
+    - Daily video transport
+    - Speech-to-text and text-to-speech services
+    - Language model integration
+    - Mem0 memory service (using either API or local configuration)
+    - RTVI event handling
+    """
+    # Note: You can pass the user_id as a parameter in API call
+    USER_ID = "pipecat-demo-user"
+
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    # Initialize text-to-speech service
+    tts = ElevenLabsTTSService(
+        api_key=os.getenv("ELEVENLABS_API_KEY"),
+        settings=ElevenLabsTTSService.Settings(
+            voice="pNInz6obpgDQGcFmaJgB",
+        ),
+    )
+
+    # =====================================================================
+    # OPTION 1: Using Mem0 API (cloud-based approach)
+    # This approach uses Mem0's cloud service for memory management
+    # Requires: MEM0_API_KEY set in your environment
+    # =====================================================================
+    memory = Mem0MemoryService(
+        api_key=os.getenv("MEM0_API_KEY"),  # Your Mem0 API key
+        user_id=USER_ID,  # Unique identifier for the user
+        agent_id="agent1",  # Optional identifier for the agent
+        run_id="session1",  # Optional identifier for the run
+        params=Mem0MemoryService.InputParams(
+            search_limit=10,
+            search_threshold=0.3,
+            api_version="v2",
+            system_prompt="Based on previous conversations, I recall: \n\n",
+            add_as_system_message=True,
+            position=1,
+        ),
+    )
+
+    # =====================================================================
+    # OPTION 2: Using Mem0 with local configuration (self-hosted approach)
+    # This approach uses a local LLM configuration for memory management
+    # Requires: Anthropic API key if using Claude model
+    # =====================================================================
+    # Uncomment the following code and comment out the previous memory initialization to use local config
+
+    # local_config = {
+    #     "llm": {
+    #         "provider": "anthropic",
+    #         "config": {
+    #             "model": "claude-3-5-sonnet-20240620",
+    #             "api_key": os.getenv("ANTHROPIC_API_KEY"),  # Make sure to set this in your .env
+    #         }
+    #     },
+    #     "embedder": {
+    #         "provider": "openai",
+    #         "config": {
+    #             "model": "text-embedding-3-large"
+    #         }
+    #     }
+    # }
+
+    # # Initialize Mem0 memory service with local configuration
+    # memory = Mem0MemoryService(
+    #     local_config=local_config,  # Use local LLM for memory processing
+    #     user_id=USER_ID,            # Unique identifier for the user
+    #     # agent_id="agent1",        # Optional identifier for the agent
+    #     # run_id="session1",        # Optional identifier for the run
+    # )
+
+    # Initialize LLM service
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="""You are a personal assistant. You can remember things about the person you are talking to.
+                        Some Guidelines:
+                        - Make sure your responses are friendly yet short and concise.
+                        - If the user asks you to remember something, make sure to remember it.
+                        - Greet the user by their name if you know about it.
+                    """,
+        ),
+    )
+
+    # Set up conversation context and management
+    # The context_aggregator will automatically collect conversation context
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            memory,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Get personalized greeting based on user memories
+        greeting = await get_initial_greeting(memory)
+
+        # Add the greeting as an assistant message to start the conversation
+        context.add_message({"role": "developer", "content": greeting})
+
+        # Queue the context frame to start the conversation
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/observer.py

@@ -0,0 +1,185 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import (
+    BotStartedSpeakingFrame,
+    BotStoppedSpeakingFrame,
+    EndFrame,
+    InterruptionFrame,
+    LLMRunFrame,
+    TTSTextFrame,
+    UserStartedSpeakingFrame,
+)
+from pipecat.observers.base_observer import BaseObserver, FramePushed
+from pipecat.observers.loggers.debug_log_observer import DebugLogObserver, FrameEndpoint
+from pipecat.observers.loggers.llm_log_observer import LLMLogObserver
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.frame_processor import FrameDirection
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_input import BaseInputTransport
+from pipecat.transports.base_output import BaseOutputTransport
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+class CustomObserver(BaseObserver):
+    """Observer to log interruptions and bot speaking events to the console.
+
+    Logs all frame instances of:
+    - InterruptionFrame
+    - BotStartedSpeakingFrame
+    - BotStoppedSpeakingFrame
+
+    This allows you to see the frame flow from processor to processor through the pipeline for these frames.
+    Log format: [EVENT TYPE]: [source processor] → [destination processor] at [timestamp]s
+    """
+
+    async def on_push_frame(self, data: FramePushed):
+        src = data.source
+        dst = data.destination
+        frame = data.frame
+        direction = data.direction
+        timestamp = data.timestamp
+
+        # Convert timestamp to seconds for readability
+        time_sec = timestamp / 1_000_000_000
+
+        # Create direction arrow
+        arrow = "→" if direction == FrameDirection.DOWNSTREAM else "←"
+
+        if isinstance(frame, InterruptionFrame) and isinstance(src, BaseOutputTransport):
+            logger.info(f"⚡ INTERRUPTION START: {src} {arrow} {dst} at {time_sec:.2f}s")
+        elif isinstance(frame, BotStartedSpeakingFrame):
+            logger.info(f"🤖 BOT START SPEAKING: {src} {arrow} {dst} at {time_sec:.2f}s")
+        elif isinstance(frame, BotStoppedSpeakingFrame):
+            logger.info(f"🤖 BOT STOP SPEAKING: {src} {arrow} {dst} at {time_sec:.2f}s")
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+        observers=[
+            CustomObserver(),
+            LLMLogObserver(),
+            DebugLogObserver(
+                frame_types={
+                    TTSTextFrame: (BaseOutputTransport, FrameEndpoint.SOURCE),
+                    UserStartedSpeakingFrame: (BaseInputTransport, FrameEndpoint.SOURCE),
+                    EndFrame: None,
+                }
+            ),
+        ],
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/pattern-pair-voice-switching.py

@@ -0,0 +1,246 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""Pattern Pair Voice Switching Example with Pipecat.
+
+This example demonstrates how to use the PatternPairAggregator to dynamically switch
+between different voices in a storytelling application. It showcases how pattern matching
+can be used to control TTS behavior in streaming text from an LLM.
+
+The example:
+    1. Sets up a storytelling bot with three distinct voices (narrator, male, female)
+    2. Uses pattern pairs (<voice>name</voice>) to trigger voice switching
+    3. Processes the patterns in real-time as text streams from the LLM
+    4. Removes the pattern tags before sending text to TTS
+
+The PatternPairAggregator:
+    - Buffers text until complete patterns are detected
+    - Identifies content between start/end pattern pairs
+    - Triggers callbacks when patterns are matched
+    - Processes patterns that may span across multiple text chunks
+    - Returns processed text at sentence boundaries
+
+Requirements:
+    - OpenAI API key
+    - Cartesia API key (for text-to-speech)
+    - Daily API key (for video/audio transport)
+
+    Environment variables (.env file):
+        OPENAI_API_KEY=your_openai_key
+        CARTESIA_API_KEY=your_cartesia_key
+        DAILY_API_KEY=your_daily_key
+
+Note:
+    This example shows one application of PatternPairAggregator (voice switching),
+    but the same approach can be used for various pattern-based text processing needs,
+    such as formatting instructions, command recognition, or structured data extraction.
+"""
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.utils.text.pattern_pair_aggregator import (
+    MatchAction,
+    PatternMatch,
+    PatternPairAggregator,
+)
+
+load_dotenv(override=True)
+
+
+# Define voice IDs
+VOICE_IDS = {
+    "narrator": "c45bc5ec-dc68-4feb-8829-6e6b2748095d",  # Narrator voice
+    "female": "71a7ad14-091c-4e8e-a314-022ece01c121",  # Female character voice
+    "male": "7cf0e2b1-8daf-4fe4-89ad-f6039398f359",  # Male character voice
+}
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    # Create pattern pair aggregator for voice switching
+    pattern_aggregator = PatternPairAggregator()
+
+    # Add pattern for voice switching
+    pattern_aggregator.add_pattern(
+        type="voice",
+        start_pattern="<voice>",
+        end_pattern="</voice>",
+        action=MatchAction.REMOVE,  # Remove tags from final text
+    )
+
+    # Register handler for voice switching
+    async def on_voice_tag(match: PatternMatch):
+        voice_name = match.text.strip().lower()
+        if voice_name in VOICE_IDS:
+            # First flush any existing audio to finish the current context
+            await tts.flush_audio()
+            # Then set the new voice
+            await tts.set_voice(VOICE_IDS[voice_name])
+            logger.info(f"Switched to {voice_name} voice")
+        else:
+            logger.warning(f"Unknown voice: {voice_name}")
+
+    pattern_aggregator.on_pattern_match("voice", on_voice_tag)
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    # Initialize TTS with narrator voice as default
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice=VOICE_IDS["narrator"],
+        ),
+        text_aggregator=pattern_aggregator,
+    )
+
+    # System prompt for storytelling with voice switching
+    system_prompt = """You are an engaging storyteller that uses different voices to bring stories to life.
+
+You have three voices to use, but each has a specific purpose:
+
+<voice>narrator</voice>
+This is the default narrator voice. Use this for all narration, descriptions, and non-dialogue text.
+
+<voice>female</voice>
+Use this ONLY for direct speech by female characters (just the quoted text).
+
+<voice>male</voice>
+Use this ONLY for direct speech by male characters (just the quoted text).
+
+IMPORTANT: Switch back to narrator voice immediately after character dialogue.
+
+Here's an EXAMPLE of correct voice usage:
+
+<voice>narrator</voice>
+Sarah spotted her old friend across the café. She couldn't believe her eyes.
+
+<voice>female</voice>
+"Jacob! It's been so long!"
+
+<voice>narrator</voice>
+Sarah exclaimed, jumping up from her seat with a radiant smile.
+
+<voice>male</voice>
+"Sarah, is it really you? I can't believe it!"
+
+<voice>narrator</voice>
+Jacob replied, grinning widely as he walked over to her. The two friends embraced warmly, as if trying to make up for all the years spent apart.
+
+<voice>female</voice>
+"What are you doing in town? Last I heard you were in Seattle."
+
+<voice>narrator</voice>
+She asked, gesturing for him to join her at the table.
+
+FOLLOW THESE RULES:
+1. Always begin with the narrator voice
+2. Only use character voices for the EXACT words they speak (in quotes)
+3. SWITCH BACK to narrator voice for speech tags and all other text
+4. Begin by asking what kind of story the user would like to hear
+5. Create engaging dialogue with distinct characters
+
+Remember: Use narrator voice for EVERYTHING except the actual quoted dialogue."""
+
+    # Initialize LLM
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction=system_prompt,
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    # Create pipeline
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,  # TTS with pattern aggregator
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Start conversation - empty prompt to let LLM follow system instructions
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/sentry-metrics.py

@@ -0,0 +1,138 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+
+import sentry_sdk
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.metrics.sentry import SentryMetrics
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    # Initialize Sentry
+    sentry_sdk.init(
+        dsn=os.getenv("SENTRY_DSN"),
+        traces_sample_rate=1.0,
+    )
+
+    stt = DeepgramSTTService(
+        api_key=os.getenv("DEEPGRAM_API_KEY"),
+        metrics=SentryMetrics(),
+    )
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+        metrics=SentryMetrics(),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        metrics=SentryMetrics(),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/service-switcher.py

@@ -0,0 +1,200 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import asyncio
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame, ManuallySwitchServiceFrame
+from pipecat.pipeline.llm_switcher import LLMSwitcher
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.service_switcher import ServiceSwitcher
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.stt import CartesiaSTTService
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.deepgram.tts import DeepgramTTSService
+from pipecat.services.google.llm import GoogleLLMService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+# "Classic" function
+async def fetch_weather_from_api(params: FunctionCallParams):
+    await params.result_callback({"conditions": "nice", "temperature": "75"})
+
+
+# "Direct" function
+async def get_restaurant_recommendation(params: FunctionCallParams, location: str):
+    """
+    Get a restaurant recommendation.
+
+    Args:
+        location (str): The city and state, e.g. "San Francisco, CA".
+    """
+    await params.result_callback({"name": "The Golden Dragon"})
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    weather_function = FunctionSchema(
+        name="get_current_weather",
+        description="Get the current weather",
+        properties={
+            "location": {
+                "type": "string",
+                "description": "The city and state, e.g. San Francisco, CA",
+            },
+            "format": {
+                "type": "string",
+                "enum": ["celsius", "fahrenheit"],
+                "description": "The temperature unit to use. Infer this from the user's location.",
+            },
+        },
+        required=["location", "format"],
+    )
+
+    stt_cartesia = CartesiaSTTService(api_key=os.getenv("CARTESIA_API_KEY"))
+    stt_deepgram = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+    # Uses ServiceSwitcherStrategyManual by default
+    stt_switcher = ServiceSwitcher(services=[stt_cartesia, stt_deepgram])
+
+    tts_cartesia = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+    tts_deepgram = DeepgramTTSService(
+        api_key=os.getenv("DEEPGRAM_API_KEY"),
+        settings=DeepgramTTSService.Settings(
+            voice="aura-2-helena-en",
+        ),
+    )
+    # Uses ServiceSwitcherStrategyManual by default
+    tts_switcher = ServiceSwitcher(services=[tts_cartesia, tts_deepgram])
+
+    system_prompt = "You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way."
+
+    llm_openai = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(system_instruction=system_prompt),
+    )
+    llm_google = GoogleLLMService(
+        api_key=os.getenv("GOOGLE_API_KEY"),
+        settings=GoogleLLMService.Settings(system_instruction=system_prompt),
+    )
+    # Uses ServiceSwitcherStrategyManual by default
+    llm_switcher = LLMSwitcher(llms=[llm_openai, llm_google])
+    # Register a "classic" function
+    llm_switcher.register_function("get_current_weather", fetch_weather_from_api)
+    # Register a "direct" function
+    llm_switcher.register_direct_function(get_restaurant_recommendation)
+
+    tools = ToolsSchema(standard_tools=[weather_function, get_restaurant_recommendation])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt_switcher,
+            user_aggregator,  # User responses
+            llm_switcher,  # LLM
+            tts_switcher,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+        await asyncio.sleep(15)
+        print(f"Switching to {stt_deepgram}")
+        await task.queue_frames([ManuallySwitchServiceFrame(service=stt_deepgram)])
+        await asyncio.sleep(15)
+        print(f"Switching to {llm_google}")
+        await task.queue_frames([ManuallySwitchServiceFrame(service=llm_google)])
+        await asyncio.sleep(15)
+        print(f"Switching to {tts_deepgram}")
+        await task.queue_frames([ManuallySwitchServiceFrame(service=tts_deepgram)])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/sound-effects.py

@@ -0,0 +1,178 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+import wave
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import (
+    Frame,
+    LLMContextFrame,
+    LLMFullResponseEndFrame,
+    OutputAudioRawFrame,
+    TTSSpeakFrame,
+)
+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 LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.frame_processor import FrameDirection, FrameProcessor
+from pipecat.processors.logger import FrameLogger
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+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] = OutputAudioRawFrame(
+            audio_file.readframes(-1), audio_file.getframerate(), audio_file.getnchannels()
+        )
+
+
+class OutboundSoundEffectWrapper(FrameProcessor):
+    async def process_frame(self, frame: Frame, direction: FrameDirection):
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, LLMFullResponseEndFrame):
+            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):
+        await super().process_frame(frame, direction)
+
+        if isinstance(frame, LLMContextFrame):
+            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)
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+    out_sound = OutboundSoundEffectWrapper()
+    in_sound = InboundSoundEffectWrapper()
+    fl = FrameLogger("LLM Out")
+    fl2 = FrameLogger("Transcription In")
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            in_sound,
+            fl2,
+            llm,
+            fl,
+            tts,
+            out_sound,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        await task.queue_frame(TTSSpeakFrame("Hi, I'm listening!"))
+        await transport.send_audio(sounds["ding1.wav"])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/switch-languages.py

@@ -0,0 +1,190 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import Frame, LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.filters.function_filter import FunctionFilter
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+class SwitchLanguage(ParallelPipeline):
+    def __init__(self):
+        self._current_language = "English"
+
+        english_tts = CartesiaTTSService(
+            api_key=os.getenv("CARTESIA_API_KEY"),
+            settings=CartesiaTTSService.Settings(
+                voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            ),
+        )
+
+        spanish_tts = CartesiaTTSService(
+            api_key=os.getenv("CARTESIA_API_KEY"),
+            settings=CartesiaTTSService.Settings(
+                voice="d4db5fb9-f44b-4bd1-85fa-192e0f0d75f9",  # Spanish-speaking Lady
+            ),
+        )
+
+        super().__init__(
+            # English
+            [FunctionFilter(self.english_filter), english_tts],
+            # Spanish
+            [FunctionFilter(self.spanish_filter), spanish_tts],
+        )
+
+    @property
+    def current_language(self):
+        return self._current_language
+
+    async def switch_language(self, params: FunctionCallParams):
+        self._current_language = params.arguments["language"]
+        await params.result_callback(
+            {"voice": f"Your answers from now on should be in {self.current_language}."}
+        )
+
+    async def english_filter(self, _: Frame) -> bool:
+        return self.current_language == "English"
+
+    async def spanish_filter(self, _: Frame) -> bool:
+        return self.current_language == "Spanish"
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(
+        api_key=os.getenv("DEEPGRAM_API_KEY"),
+        settings=DeepgramSTTService.Settings(
+            language="multi",
+        ),
+    )
+
+    tts = SwitchLanguage()
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way. You can speak the following languages: 'English' and 'Spanish'.",
+        ),
+    )
+    llm.register_function("switch_language", tts.switch_language)
+
+    switch_language_function = FunctionSchema(
+        name="switch_language",
+        description="Switch to another language when the user asks you to",
+        properties={
+            "language": {
+                "type": "string",
+                "description": "The language the user wants you to speak",
+            },
+        },
+        required=["language"],
+    )
+    tools = ToolsSchema(standard_tools=[switch_language_function])
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,  # STT
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS (bot will speak the chosen language)
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {
+                "role": "developer",
+                "content": f"Please introduce yourself to the user and let them know the languages you speak. Your initial responses should be in {tts.current_language}.",
+            }
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/switch-voices.py

@@ -0,0 +1,200 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import Frame, LLMRunFrame
+from pipecat.pipeline.parallel_pipeline import ParallelPipeline
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.processors.filters.function_filter import FunctionFilter
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+class SwitchVoices(ParallelPipeline):
+    def __init__(self):
+        self._current_voice = "News Lady"
+
+        news_lady = CartesiaTTSService(
+            api_key=os.getenv("CARTESIA_API_KEY"),
+            settings=CartesiaTTSService.Settings(
+                voice="bf991597-6c13-47e4-8411-91ec2de5c466",  # Newslady
+            ),
+        )
+
+        british_lady = CartesiaTTSService(
+            api_key=os.getenv("CARTESIA_API_KEY"),
+            settings=CartesiaTTSService.Settings(
+                voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+            ),
+        )
+
+        barbershop_man = CartesiaTTSService(
+            api_key=os.getenv("CARTESIA_API_KEY"),
+            settings=CartesiaTTSService.Settings(
+                voice="a0e99841-438c-4a64-b679-ae501e7d6091",  # Barbershop Man
+            ),
+        )
+
+        super().__init__(
+            # News Lady voice
+            [FunctionFilter(self.news_lady_filter), news_lady],
+            # British Reading Lady voice
+            [FunctionFilter(self.british_lady_filter), british_lady],
+            # Barbershop Man voice
+            [FunctionFilter(self.barbershop_man_filter), barbershop_man],
+        )
+
+    @property
+    def current_voice(self):
+        return self._current_voice
+
+    async def switch_voice(self, params: FunctionCallParams):
+        self._current_voice = params.arguments["voice"]
+        await params.result_callback(
+            {
+                "voice": f"You are now using your {self.current_voice} voice. Your responses should now be as if you were a {self.current_voice}."
+            }
+        )
+
+    async def news_lady_filter(self, _: Frame) -> bool:
+        return self.current_voice == "News Lady"
+
+    async def british_lady_filter(self, _: Frame) -> bool:
+        return self.current_voice == "British Lady"
+
+    async def barbershop_man_filter(self, _: Frame) -> bool:
+        return self.current_voice == "Barbershop Man"
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = SwitchVoices()
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative and helpful way. You can do the following voices: 'News Lady', 'British Lady' and 'Barbershop Man'.",
+        ),
+    )
+    llm.register_function("switch_voice", tts.switch_voice)
+
+    switch_voice_function = FunctionSchema(
+        name="switch_voice",
+        description="Switch your voice only when the user asks you to",
+        properties={
+            "voice": {
+                "type": "string",
+                "description": "The voice the user wants you to use",
+            },
+        },
+        required=["voice"],
+    )
+    tools = ToolsSchema(standard_tools=[switch_voice_function])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS with switch voice functionality
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {
+                "role": "developer",
+                "content": f"Please introduce yourself to the user and let them know the voices you can do. Your initial responses should be as if you were a {tts.current_voice}.",
+            }
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/user-email-gathering.py

@@ -0,0 +1,161 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.adapters.schemas.function_schema import FunctionSchema
+from pipecat.adapters.schemas.tools_schema import ToolsSchema
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.llm_service import FunctionCallParams
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+
+async def store_user_emails(params: FunctionCallParams):
+    print(f"User emails: {params.arguments}")
+
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    # Cartesia offers a `<spell></spell>` tags that we can use to ask the user
+    # to confirm the emails.
+    # (see https://docs.cartesia.ai/build-with-sonic/formatting-text-for-sonic/spelling-out-input-text)
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    # Rime offers a function `spell()` that we can use to ask the user
+    # to confirm the emails.
+    # (see https://docs.rime.ai/api-reference/spell)
+    # tts = RimeHttpTTSService(
+    #     api_key=os.getenv("RIME_API_KEY", ""),
+    #     settings=RimeTTSSettings(
+    #         voice="eva",
+    #     ),
+    #     aiohttp_session=session,
+    # )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You need to gather a valid email or emails from the user. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. If the user provides one or more email addresses confirm them with the user. Enclose all emails with <spell> tags, for example <spell>a@a.com</spell>.",
+        ),
+    )
+    # You can aslo register a function_name of None to get all functions
+    # sent to the same callback with an additional function_name parameter.
+    llm.register_function("store_user_emails", store_user_emails)
+
+    store_emails_function = FunctionSchema(
+        name="store_user_emails",
+        description="Store user emails when confirmed",
+        properties={
+            "emails": {
+                "type": "array",
+                "description": "The list of user emails",
+                "items": {"type": "string"},
+            },
+        },
+        required=["emails"],
+    )
+    tools = ToolsSchema(standard_tools=[store_emails_function])
+
+    context = LLMContext(tools=tools)
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            user_aggregator,
+            llm,
+            tts,
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Start conversation - empty prompt to let LLM follow system instructions
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/voicemail-detection.py

@@ -0,0 +1,145 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.extensions.voicemail.voicemail_detector import VoicemailDetector
+from pipecat.frames.frames import TTSSpeakFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(api_key=os.getenv("DEEPGRAM_API_KEY"))
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+    classifier_llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
+
+    voicemail = VoicemailDetector(llm=classifier_llm)
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(vad_analyzer=SileroVADAnalyzer()),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),
+            stt,
+            voicemail.detector(),  # Voicemail detection — between STT and User context aggregator
+            user_aggregator,
+            llm,
+            tts,
+            voicemail.gate(),  # TTS gating — Immediately after the TTS service
+            transport.output(),
+            assistant_aggregator,
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    @voicemail.event_handler("on_conversation_detected")
+    async def on_conversation_detected(processor):
+        logger.info("Conversation detected!")
+
+    @voicemail.event_handler("on_voicemail_detected")
+    async def on_voicemail_detected(processor):
+        logger.info("Voicemail detected! Leaving a message...")
+
+        # Push frames using standard Pipecat pattern
+        await processor.push_frame(
+            TTSSpeakFrame(
+                "Hello, this is Jamie calling about your appointment. Please call me back at 555-0123 when you get this."
+            )
+        )
+
+        # NOTE: A common pattern is to end pipeline after the voicemail is left.
+        # Uncomment the following line to end the pipeline after leaving the voicemail.
+        # await processor.push_frame(EndTaskFrame(), FrameDirection.UPSTREAM)
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()

examples/features/wake-phrase.py

@@ -0,0 +1,147 @@
+#
+# Copyright (c) 2024-2026, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+import os
+
+from dotenv import load_dotenv
+from loguru import logger
+
+from pipecat.audio.vad.silero import SileroVADAnalyzer
+from pipecat.frames.frames import LLMRunFrame
+from pipecat.pipeline.pipeline import Pipeline
+from pipecat.pipeline.runner import PipelineRunner
+from pipecat.pipeline.task import PipelineParams, PipelineTask
+from pipecat.processors.aggregators.llm_context import LLMContext
+from pipecat.processors.aggregators.llm_response_universal import (
+    LLMContextAggregatorPair,
+    LLMUserAggregatorParams,
+)
+from pipecat.runner.types import RunnerArguments
+from pipecat.runner.utils import create_transport
+from pipecat.services.cartesia.tts import CartesiaTTSService
+from pipecat.services.deepgram.stt import DeepgramSTTService
+from pipecat.services.openai.llm import OpenAILLMService
+from pipecat.transports.base_transport import BaseTransport, TransportParams
+from pipecat.transports.daily.transport import DailyParams
+from pipecat.transports.websocket.fastapi import FastAPIWebsocketParams
+from pipecat.turns.user_start import WakePhraseUserTurnStartStrategy
+from pipecat.turns.user_turn_strategies import (
+    UserTurnStrategies,
+    default_user_turn_start_strategies,
+)
+
+load_dotenv(override=True)
+
+# We use lambdas to defer transport parameter creation until the transport
+# type is selected at runtime.
+transport_params = {
+    "daily": lambda: DailyParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "twilio": lambda: FastAPIWebsocketParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+    "webrtc": lambda: TransportParams(
+        audio_in_enabled=True,
+        audio_out_enabled=True,
+    ),
+}
+
+
+async def run_bot(transport: BaseTransport, runner_args: RunnerArguments):
+    logger.info(f"Starting bot")
+
+    stt = DeepgramSTTService(
+        api_key=os.getenv("DEEPGRAM_API_KEY"),
+        settings=DeepgramSTTService.Settings(
+            keyterm=["pipecat"],
+        ),
+    )
+
+    tts = CartesiaTTSService(
+        api_key=os.getenv("CARTESIA_API_KEY"),
+        settings=CartesiaTTSService.Settings(
+            voice="71a7ad14-091c-4e8e-a314-022ece01c121",  # British Reading Lady
+        ),
+    )
+
+    llm = OpenAILLMService(
+        api_key=os.getenv("OPENAI_API_KEY"),
+        settings=OpenAILLMService.Settings(
+            system_instruction="You are a helpful assistant in a voice conversation. Your responses will be spoken aloud, so avoid emojis, bullet points, or other formatting that can't be spoken. Respond to what the user said in a creative, helpful, and brief way.",
+        ),
+    )
+
+    context = LLMContext()
+    user_aggregator, assistant_aggregator = LLMContextAggregatorPair(
+        context,
+        user_params=LLMUserAggregatorParams(
+            user_turn_strategies=UserTurnStrategies(
+                start=[
+                    WakePhraseUserTurnStartStrategy(
+                        phrases=["pipecat"],
+                        # Timeout before wake phrase must be spoken again
+                        timeout=5.0,
+                    ),
+                    *default_user_turn_start_strategies(),
+                ]
+            ),
+            vad_analyzer=SileroVADAnalyzer(),
+        ),
+    )
+
+    pipeline = Pipeline(
+        [
+            transport.input(),  # Transport user input
+            stt,
+            user_aggregator,  # User responses
+            llm,  # LLM
+            tts,  # TTS
+            transport.output(),  # Transport bot output
+            assistant_aggregator,  # Assistant spoken responses
+        ]
+    )
+
+    task = PipelineTask(
+        pipeline,
+        params=PipelineParams(
+            enable_metrics=True,
+            enable_usage_metrics=True,
+        ),
+        idle_timeout_secs=runner_args.pipeline_idle_timeout_secs,
+    )
+
+    @transport.event_handler("on_client_connected")
+    async def on_client_connected(transport, client):
+        logger.info(f"Client connected")
+        # Kick off the conversation.
+        context.add_message(
+            {"role": "developer", "content": "Please introduce yourself to the user."}
+        )
+        await task.queue_frames([LLMRunFrame()])
+
+    @transport.event_handler("on_client_disconnected")
+    async def on_client_disconnected(transport, client):
+        logger.info(f"Client disconnected")
+        await task.cancel()
+
+    runner = PipelineRunner(handle_sigint=runner_args.handle_sigint)
+
+    await runner.run(task)
+
+
+async def bot(runner_args: RunnerArguments):
+    """Main bot entry point compatible with Pipecat Cloud."""
+    transport = await create_transport(runner_args, transport_params)
+    await run_bot(transport, runner_args)
+
+
+if __name__ == "__main__":
+    from pipecat.runner.run import main
+
+    main()