wx44wx/pipecat
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Update Smallest AI services to use ServiceSettings and add examples

Browse Source 
Migrate STT/TTS services from deprecated set_model_name()/set_voice() to the
new ServiceSettings pattern (STTSettings/TTSSettings). Add default voice_id
("sophia") for TTS services, fix voice references, and include two foundational
example scripts showing WebSocket and HTTP usage.

Made-with: Cursor
This commit is contained in:
Harshita Jain
2026-02-27 16:10:49 -08:00
committed by Mark Backman
parent e62b416056
commit 40c36f8a2a
4 changed files with 290 additions and 47 deletions
Download Patch File Download Diff File Expand all files Collapse all files

122
examples/foundational/07zl-interruptible-smallest-http.py Normal file
View File

@@ -0,0 +1,122 @@
#
# 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.openai.llm import OpenAILLMService
from pipecat.services.smallest.stt import SmallestSTTService
from pipecat.services.smallest.tts import SmallestHttpTTSService
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)
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 = SmallestSTTService(
api_key=os.getenv("SMALLEST_API_KEY"),
)
tts = SmallestHttpTTSService(
api_key=os.getenv("SMALLEST_API_KEY"),
voice_id="sophia",
)
llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
messages = [
{
"role": "system",
"content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
},
]
context = LLMContext(messages)
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,
),
)
@transport.event_handler("on_client_connected")
async def on_client_connected(transport, client):
logger.info(f"Client connected")
messages.append({"role": "system", "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()

122
examples/foundational/07zl-interruptible-smallest.py Normal file
View File

@@ -0,0 +1,122 @@
#
# 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.openai.llm import OpenAILLMService
from pipecat.services.smallest.stt import SmallestRealtimeSTTService
from pipecat.services.smallest.tts import SmallestTTSService
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)
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 = SmallestRealtimeSTTService(
api_key=os.getenv("SMALLEST_API_KEY"),
)
tts = SmallestTTSService(
api_key=os.getenv("SMALLEST_API_KEY"),
voice_id="sophia",
)
llm = OpenAILLMService(api_key=os.getenv("OPENAI_API_KEY"))
messages = [
{
"role": "system",
"content": "You are a helpful LLM in a WebRTC call. Your goal is to demonstrate your capabilities in a succinct way. Your output will be spoken aloud, so avoid special characters that can't easily be spoken, such as emojis or bullet points. Respond to what the user said in a creative and helpful way.",
},
]
context = LLMContext(messages)
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,
),
)
@transport.event_handler("on_client_connected")
async def on_client_connected(transport, client):
logger.info(f"Client connected")
messages.append({"role": "system", "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()

34
src/pipecat/services/smallest/stt.py
View File

@@ -36,6 +36,7 @@ from pipecat.frames.frames import (
VADUserStoppedSpeakingFrame,
)
from pipecat.processors.frame_processor import FrameDirection
from pipecat.services.settings import STTSettings
from pipecat.services.stt_latency import SMALLEST_TTFS_P99
from pipecat.services.stt_service import SegmentedSTTService, WebsocketSTTService
from pipecat.transcriptions.language import Language
@@ -154,21 +155,20 @@ class SmallestSTTService(SegmentedSTTService):
Override for your deployment.
**kwargs: Additional arguments passed to the parent SegmentedSTTService.
"""
params = params or SmallestSTTService.InputParams()
model_str = model.value if isinstance(model, Enum) else model
super().__init__(
sample_rate=sample_rate,
ttfs_p99_latency=ttfs_p99_latency,
settings=STTSettings(model=model_str, language=params.language),
**kwargs,
)
params = params or SmallestSTTService.InputParams()
self._api_key = api_key
self._url = url
self._language = params.language
model_str = model.value if isinstance(model, Enum) else model
self.set_model_name(model_str)
self._client = httpx.AsyncClient()
self._headers = {
"Authorization": f"Bearer {self._api_key}",
@@ -340,23 +340,23 @@ class SmallestRealtimeSTTService(WebsocketSTTService):
ttfs_p99_latency: P99 latency from speech end to final transcript in seconds.
**kwargs: Additional arguments passed to WebsocketSTTService.
"""
self._rt_params = params or SmallestRealtimeSTTService.InputParams()
super().__init__(
sample_rate=sample_rate,
ttfs_p99_latency=ttfs_p99_latency,
keepalive_timeout=10,
keepalive_interval=5,
settings=STTSettings(model="pulse", language=self._rt_params.language),
**kwargs,
)
self._api_key = api_key
self._base_url = base_url.rstrip("/")
self._params = params or SmallestRealtimeSTTService.InputParams()
self._receive_task = None
self._connected_event = asyncio.Event()
self._connected_event.set()
self.set_model_name("pulse")
def can_generate_metrics(self) -> bool:
"""Check if this service can generate processing metrics."""
return True
@@ -442,16 +442,16 @@ class SmallestRealtimeSTTService(WebsocketSTTService):
logger.debug("Connecting to Smallest Realtime STT")
query_params = {
"language": self._params.language,
"encoding": self._params.encoding,
"language": self._rt_params.language,
"encoding": self._rt_params.encoding,
"sample_rate": str(self.sample_rate),
"word_timestamps": str(self._params.word_timestamps).lower(),
"full_transcript": str(self._params.full_transcript).lower(),
"sentence_timestamps": str(self._params.sentence_timestamps).lower(),
"redact_pii": str(self._params.redact_pii).lower(),
"redact_pci": str(self._params.redact_pci).lower(),
"numerals": self._params.numerals,
"diarize": str(self._params.diarize).lower(),
"word_timestamps": str(self._rt_params.word_timestamps).lower(),
"full_transcript": str(self._rt_params.full_transcript).lower(),
"sentence_timestamps": str(self._rt_params.sentence_timestamps).lower(),
"redact_pii": str(self._rt_params.redact_pii).lower(),
"redact_pci": str(self._rt_params.redact_pci).lower(),
"numerals": self._rt_params.numerals,
"diarize": str(self._rt_params.diarize).lower(),
}
ws_url = f"{self._base_url}/waves/v1/pulse/get_text?{urlencode(query_params)}"

59
src/pipecat/services/smallest/tts.py
View File

@@ -30,6 +30,7 @@ from pipecat.frames.frames import (
TTSStoppedFrame,
)
from pipecat.processors.frame_processor import FrameDirection
from pipecat.services.settings import TTSSettings
from pipecat.services.tts_service import InterruptibleTTSService, TTSService
from pipecat.transcriptions.language import Language
from pipecat.utils.tracing.service_decorators import traced_tts
@@ -99,7 +100,7 @@ class SmallestTTSService(InterruptibleTTSService):
tts = SmallestTTSService(
api_key="your-api-key",
voice_id="emily",
voice_id="sophia",
params=SmallestTTSService.InputParams(
language=Language.EN,
speed=1.0,
@@ -128,7 +129,7 @@ class SmallestTTSService(InterruptibleTTSService):
self,
*,
api_key: str,
voice_id: str,
voice_id: str = "sophia",
base_url: str = "wss://waves-api.smallest.ai",
model: str = SmallestTTSModel.LIGHTNING_V3_1,
sample_rate: Optional[int] = 24000,
@@ -146,27 +147,26 @@ class SmallestTTSService(InterruptibleTTSService):
params: Configuration parameters for the TTS service.
**kwargs: Additional arguments passed to parent InterruptibleTTSService.
"""
params = params or SmallestTTSService.InputParams()
model_str = model.value if isinstance(model, Enum) else model
lang_str = (
language_to_smallest_tts_language(params.language) if params.language else "en"
)
super().__init__(
aggregate_sentences=True,
push_text_frames=True,
pause_frame_processing=True,
sample_rate=sample_rate,
settings=TTSSettings(model=model_str, voice=voice_id, language=lang_str),
**kwargs,
)
params = params or SmallestTTSService.InputParams()
self._api_key = api_key
model_str = model.value if isinstance(model, Enum) else model
self._websocket_url = f"{base_url}/api/v1/{model_str}/get_speech/stream"
self.set_model_name(model_str)
self.set_voice(voice_id)
self._settings = {
"language": language_to_smallest_tts_language(params.language)
if params.language
else "en",
self._tts_params = {
"language": lang_str,
"speed": params.speed,
"consistency": params.consistency,
"similarity": params.similarity,
@@ -206,12 +206,12 @@ class SmallestTTSService(InterruptibleTTSService):
"""
msg = {
"text": text,
"voice_id": self._voice_id,
"language": self._settings["language"],
"speed": self._settings["speed"],
"consistency": self._settings["consistency"],
"similarity": self._settings["similarity"],
"enhancement": self._settings["enhancement"],
"voice_id": self._settings.voice,
"language": self._tts_params["language"],
"speed": self._tts_params["speed"],
"consistency": self._tts_params["consistency"],
"similarity": self._tts_params["similarity"],
"enhancement": self._tts_params["enhancement"],
}
if self._context_id:
@@ -431,7 +431,7 @@ class SmallestHttpTTSService(TTSService):
self,
*,
api_key: str,
voice_id: str,
voice_id: str = "sophia",
model: str = SmallestTTSModel.LIGHTNING_V3_1,
base_url: str = "https://waves-api.smallest.ai",
sample_rate: Optional[int] = None,
@@ -449,20 +449,20 @@ class SmallestHttpTTSService(TTSService):
params: Configuration parameters for the TTS service.
**kwargs: Additional arguments passed to parent TTSService.
"""
super().__init__(sample_rate=sample_rate, **kwargs)
params = params or SmallestHttpTTSService.InputParams()
model_str = model.value if isinstance(model, Enum) else model
super().__init__(
sample_rate=sample_rate,
settings=TTSSettings(model=model_str, voice=voice_id, language=params.language),
**kwargs,
)
self._api_key = api_key
self._base_url = base_url.rstrip("/")
model_str = model.value if isinstance(model, Enum) else model
self.set_model_name(model_str)
self.set_voice(voice_id)
self._model_url = f"{self._base_url}/api/v1/{model_str}/get_speech"
self._settings = {
self._tts_params = {
"language": params.language,
"speed": params.speed,
"consistency": params.consistency,
@@ -539,13 +539,12 @@ class SmallestHttpTTSService(TTSService):
await self.start_ttfb_metrics()
payload = {
"voice_id": self._voice_id,
"voice_id": self._settings.voice,
"text": text,
"sample_rate": self.sample_rate,
}
# Only include non-None settings
for key, value in self._settings.items():
for key, value in self._tts_params.items():
if value is not None:
payload[key] = value