From 099814d74a84066cc49678e07e0b36aaf8f69b68 Mon Sep 17 00:00:00 2001 From: Harshita Jain Date: Fri, 20 Mar 2026 09:30:16 -0700 Subject: [PATCH] Add Smallest AI TTS service integration Adds SmallestTTSService, a WebSocket-based TTS service using Smallest AI's Lightning v3.1 model. Follows current Pipecat service conventions: - SmallestTTSSettings dataclass with runtime-updatable settings (voice, language, speed, etc.) - Reconnects on model change; keepalive every 30s to prevent idle timeout - TTS settings default to None so the API applies its own defaults - Model enum: SmallestTTSModel.LIGHTNING_V3_1 Includes a foundational example (07zl-interruptible-smallest.py) using Deepgram STT + Smallest TTS + OpenAI LLM. STT integration will follow in a separate PR once the hallucination/finalize behaviour is resolved. Made-with: Cursor --- .../07zl-interruptible-smallest.py | 122 +++++ pyproject.toml | 1 + src/pipecat/services/smallest/__init__.py | 1 + src/pipecat/services/smallest/tts.py | 424 ++++++++++++++++++ uv.lock | 2 +- 5 files changed, 549 insertions(+), 1 deletion(-) create mode 100644 examples/foundational/07zl-interruptible-smallest.py create mode 100644 src/pipecat/services/smallest/__init__.py create mode 100644 src/pipecat/services/smallest/tts.py diff --git a/examples/foundational/07zl-interruptible-smallest.py b/examples/foundational/07zl-interruptible-smallest.py new file mode 100644 index 000000000..8ffa48245 --- /dev/null +++ b/examples/foundational/07zl-interruptible-smallest.py @@ -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.deepgram.stt import DeepgramSTTService +from pipecat.services.openai.llm import OpenAILLMService +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 = DeepgramSTTService( + api_key=os.getenv("DEEPGRAM_API_KEY"), + ) + + tts = SmallestTTSService( + api_key=os.getenv("SMALLEST_API_KEY"), + settings=SmallestTTSService.Settings( + voice="sophia", + ), + ) + + llm = OpenAILLMService( + api_key=os.getenv("OPENAI_API_KEY"), + settings=OpenAILLMService.Settings( + system_instruction="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() + 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") + context.add_message({"role": "user", "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() diff --git a/pyproject.toml b/pyproject.toml index 0416d6367..57fe00105 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -111,6 +111,7 @@ runner = [ "python-dotenv>=1.0.0,<2.0.0", "uvicorn>=0.32.0,<1.0.0", "fastapi>=0. sagemaker = ["aws_sdk_sagemaker_runtime_http2; python_version>='3.12'"] sambanova = [] sarvam = [ "sarvamai==0.1.26", "pipecat-ai[websockets-base]" ] +smallest = [ "pipecat-ai[websockets-base]" ] sentry = [ "sentry-sdk>=2.28.0,<3" ] silero = [] simli = [ "simli-ai~=2.0.1"] diff --git a/src/pipecat/services/smallest/__init__.py b/src/pipecat/services/smallest/__init__.py new file mode 100644 index 000000000..8b1378917 --- /dev/null +++ b/src/pipecat/services/smallest/__init__.py @@ -0,0 +1 @@ + diff --git a/src/pipecat/services/smallest/tts.py b/src/pipecat/services/smallest/tts.py new file mode 100644 index 000000000..ea2d03c66 --- /dev/null +++ b/src/pipecat/services/smallest/tts.py @@ -0,0 +1,424 @@ +# +# Copyright (c) 2024-2026, Daily +# +# SPDX-License-Identifier: BSD 2-Clause License +# + +"""Smallest AI text-to-speech service implementation. + +This module provides a WebSocket-based integration with Smallest AI's +Waves API for real-time text-to-speech synthesis. +""" + +import asyncio +import base64 +import json +from dataclasses import dataclass, field +from enum import Enum +from typing import Any, AsyncGenerator, Optional + +from loguru import logger + +from pipecat.frames.frames import ( + CancelFrame, + EndFrame, + ErrorFrame, + Frame, + StartFrame, + TTSAudioRawFrame, + TTSStartedFrame, + TTSStoppedFrame, +) +from pipecat.services.settings import NOT_GIVEN, TTSSettings, _NotGiven +from pipecat.services.tts_service import InterruptibleTTSService +from pipecat.transcriptions.language import Language +from pipecat.utils.tracing.service_decorators import traced_tts + +try: + from websockets.asyncio.client import connect as websocket_connect + from websockets.protocol import State +except ModuleNotFoundError as e: + logger.error(f"Exception: {e}") + logger.error("In order to use Smallest, you need to `pip install pipecat-ai[smallest]`.") + raise Exception(f"Missing module: {e}") + + +class SmallestTTSModel(str, Enum): + """Available Smallest AI TTS models.""" + + LIGHTNING_V2 = "lightning-v2" + LIGHTNING_V3_1 = "lightning-v3.1" + + +def language_to_smallest_tts_language(language: Language) -> Optional[str]: + """Convert a Language enum to a Smallest TTS language string. + + Args: + language: The Language enum value to convert. + + Returns: + The Smallest language code string, or None if unsupported. + """ + BASE_LANGUAGES = { + Language.AR: "ar", + Language.BN: "bn", + Language.DE: "de", + Language.EN: "en", + Language.ES: "es", + Language.FR: "fr", + Language.GU: "gu", + Language.HE: "he", + Language.HI: "hi", + Language.IT: "it", + Language.KN: "kn", + Language.MR: "mr", + Language.NL: "nl", + Language.PL: "pl", + Language.RU: "ru", + Language.TA: "ta", + } + + result = BASE_LANGUAGES.get(language) + + if not result: + lang_str = str(language.value) + base_code = lang_str.split("-")[0].lower() + result = base_code if base_code in BASE_LANGUAGES.values() else None + + return result + + +@dataclass +class SmallestTTSSettings(TTSSettings): + """Settings for SmallestTTSService. + + Parameters: + speed: Speech speed multiplier. + consistency: Consistency level for voice generation (0-1). + similarity: Similarity level for voice generation (0-1). + enhancement: Enhancement level for voice generation (0-2). + """ + + speed: float | None | _NotGiven = field(default_factory=lambda: NOT_GIVEN) + consistency: float | None | _NotGiven = field(default_factory=lambda: NOT_GIVEN) + similarity: float | None | _NotGiven = field(default_factory=lambda: NOT_GIVEN) + enhancement: int | None | _NotGiven = field(default_factory=lambda: NOT_GIVEN) + + +class SmallestTTSService(InterruptibleTTSService): + """Smallest AI real-time text-to-speech service using WebSocket streaming. + + Provides real-time text-to-speech synthesis using Smallest AI's WebSocket API. + Supports streaming audio generation with configurable voice parameters and + language settings. Handles interruptions by reconnecting the WebSocket. + + Example:: + + tts = SmallestTTSService( + api_key="your-api-key", + settings=SmallestTTSService.Settings( + voice="sophia", + language="en", + speed=1.0, + ), + ) + """ + + Settings = SmallestTTSSettings + _settings: Settings + + def __init__( + self, + *, + api_key: str, + base_url: str = "wss://waves-api.smallest.ai", + sample_rate: Optional[int] = None, + settings: Optional[Settings] = None, + **kwargs, + ): + """Initialize the Smallest AI WebSocket TTS service. + + Args: + api_key: Smallest AI API key for authentication. + base_url: Base WebSocket URL for the Smallest API. + sample_rate: Audio sample rate in Hz. If None, uses default. + settings: Runtime-updatable settings for the TTS service. + **kwargs: Additional arguments passed to parent InterruptibleTTSService. + """ + default_settings = self.Settings( + model=SmallestTTSModel.LIGHTNING_V3_1.value, + voice="sophia", + language=language_to_smallest_tts_language(Language.EN), + speed=None, + consistency=None, + similarity=None, + enhancement=None, + ) + + if settings is not None: + default_settings.apply_update(settings) + + super().__init__( + aggregate_sentences=True, + push_text_frames=True, + pause_frame_processing=True, + sample_rate=sample_rate, + settings=default_settings, + **kwargs, + ) + + self._api_key = api_key + self._base_url = base_url.rstrip("/") + self._receive_task = None + self._keepalive_task = None + self._context_id: Optional[str] = None + + def can_generate_metrics(self) -> bool: + """Check if this service can generate processing metrics. + + Returns: + True, as Smallest service supports metrics generation. + """ + return True + + def language_to_service_language(self, language: Language) -> Optional[str]: + """Convert a Language enum to Smallest service language format. + + Args: + language: The language to convert. + + Returns: + The Smallest-specific language code, or None if not supported. + """ + return language_to_smallest_tts_language(language) + + def _build_msg(self, text: str) -> dict: + """Build a WebSocket message for the Smallest API. + + Args: + text: The text to synthesize. + + Returns: + Dictionary with the API message payload. + """ + msg = { + "text": text, + "voice_id": self._settings.voice, + "language": self._settings.language, + "sample_rate": self.sample_rate, + } + + if self._settings.speed is not None: + msg["speed"] = self._settings.speed + if self._settings.consistency is not None: + msg["consistency"] = self._settings.consistency + if self._settings.similarity is not None: + msg["similarity"] = self._settings.similarity + if self._settings.enhancement is not None: + msg["enhancement"] = self._settings.enhancement + + if self._context_id: + msg["request_id"] = self._context_id + + return msg + + def _build_websocket_url(self) -> str: + """Build the WebSocket URL from base URL and model.""" + return f"{self._base_url}/api/v1/{self._settings.model}/get_speech/stream" + + async def start(self, frame: StartFrame): + """Start the Smallest TTS service. + + Args: + frame: The start frame containing initialization parameters. + """ + await super().start(frame) + await self._connect() + + async def stop(self, frame: EndFrame): + """Stop the Smallest TTS service. + + Args: + frame: The end frame. + """ + await super().stop(frame) + await self._disconnect() + + async def cancel(self, frame: CancelFrame): + """Cancel the Smallest TTS service. + + Args: + frame: The cancel frame. + """ + await super().cancel(frame) + await self._disconnect() + + async def _update_settings(self, delta: TTSSettings) -> dict[str, Any]: + """Apply a settings delta, reconnecting if model changed. + + Per-message fields (speed, consistency, similarity, enhancement, voice, + language) apply automatically on the next ``_build_msg`` call. A model + change requires reconnecting because the model is part of the WebSocket URL. + """ + changed = await super()._update_settings(delta) + + if not changed: + return changed + + if "model" in changed: + await self._disconnect() + await self._connect() + + return changed + + async def _connect(self): + """Connect to Smallest WebSocket and start receive task.""" + await super()._connect() + + await self._connect_websocket() + + if self._websocket and not self._receive_task: + self._receive_task = self.create_task(self._receive_task_handler(self._report_error)) + + if self._websocket and not self._keepalive_task: + self._keepalive_task = self.create_task(self._keepalive_task_handler()) + + async def _disconnect(self): + """Disconnect from Smallest WebSocket and clean up tasks.""" + await super()._disconnect() + + if self._receive_task: + await self.cancel_task(self._receive_task) + self._receive_task = None + + if self._keepalive_task: + await self.cancel_task(self._keepalive_task) + self._keepalive_task = None + + await self._disconnect_websocket() + + async def _connect_websocket(self): + """Establish WebSocket connection to the Smallest API.""" + try: + if self._websocket and self._websocket.state is State.OPEN: + return + + logger.debug("Connecting to Smallest TTS") + + self._websocket = await websocket_connect( + self._build_websocket_url(), + additional_headers={"Authorization": f"Bearer {self._api_key}"}, + ) + + await self._call_event_handler("on_connected") + except Exception as e: + await self.push_error(error_msg=f"Smallest TTS connection error: {e}", exception=e) + self._websocket = None + await self._call_event_handler("on_connection_error", f"{e}") + + async def _disconnect_websocket(self): + """Close the WebSocket connection and clean up state.""" + try: + await self.stop_all_metrics() + + if self._websocket: + logger.debug("Disconnecting from Smallest TTS") + await self._websocket.close() + except Exception as e: + logger.error(f"{self} error closing websocket: {e}") + finally: + self._context_id = None + self._websocket = None + await self._call_event_handler("on_disconnected") + + def _get_websocket(self): + """Get the WebSocket connection if available. + + Returns: + The active WebSocket connection. + + Raises: + Exception: If no WebSocket connection is available. + """ + if self._websocket: + return self._websocket + raise Exception("Websocket not connected") + + async def _keepalive_task_handler(self): + """Send periodic keepalive messages to prevent idle timeout.""" + KEEPALIVE_INTERVAL = 30 + while True: + await asyncio.sleep(KEEPALIVE_INTERVAL) + await self._send_keepalive() + + async def _send_keepalive(self): + """Send a flush message to keep the connection alive.""" + if self._websocket and self._websocket.state is State.OPEN: + msg = {"flush": True} + await self._websocket.send(json.dumps(msg)) + + async def _receive_messages(self): + """Receive and process messages from the Smallest WebSocket API.""" + async for message in self._get_websocket(): + msg = json.loads(message) + status = msg.get("status") + + if status == "complete": + msg_request_id = msg.get("request_id") + if self._context_id and msg_request_id and msg_request_id == self._context_id: + await self.stop_all_metrics() + await self.push_frame(TTSStoppedFrame(context_id=self._context_id)) + self._context_id = None + elif status == "chunk": + await self.stop_ttfb_metrics() + frame = TTSAudioRawFrame( + audio=base64.b64decode(msg["data"]["audio"]), + sample_rate=self.sample_rate, + num_channels=1, + context_id=self._context_id, + ) + await self.push_frame(frame) + elif status == "error": + logger.error(f"{self} error: {msg}") + await self.push_frame(TTSStoppedFrame(context_id=self._context_id)) + await self.stop_all_metrics() + await self.push_error(error_msg=f"Smallest TTS error: {msg.get('error', msg)}") + self._context_id = None + else: + logger.warning(f"{self} unknown message status: {msg}") + + @traced_tts + async def run_tts(self, text: str, context_id: str) -> AsyncGenerator[Frame, None]: + """Generate speech from text using Smallest's WebSocket streaming API. + + Args: + text: The text to synthesize into speech. + context_id: Unique identifier for this TTS context. + + Yields: + Frame: TTSStartedFrame to signal start; audio arrives via WebSocket. + """ + logger.debug(f"{self}: Generating TTS [{text}]") + + try: + if not self._websocket or self._websocket.state is State.CLOSED: + await self._connect() + + try: + self._context_id = context_id + yield TTSStartedFrame(context_id=context_id) + + msg = self._build_msg(text=text) + await self._get_websocket().send(json.dumps(msg)) + await self.start_tts_usage_metrics(text) + except Exception as e: + logger.error(f"{self} error sending message: {e}") + yield ErrorFrame(error=f"Smallest TTS send error: {e}") + yield TTSStoppedFrame(context_id=context_id) + await self._disconnect() + await self._connect() + return + yield None + except Exception as e: + logger.error(f"{self} exception: {e}") + yield ErrorFrame(error=f"Smallest TTS error: {e}") diff --git a/uv.lock b/uv.lock index 0089729ef..a6afce62c 100644 --- a/uv.lock +++ b/uv.lock @@ -8480,4 +8480,4 @@ wheels = [ { url = "https://files.pythonhosted.org/packages/c2/38/f249a2050ad1eea0bb364046153942e34abba95dd5520af199aed86fbb49/zstandard-0.25.0-cp314-cp314-win32.whl", hash = "sha256:da469dc041701583e34de852d8634703550348d5822e66a0c827d39b05365b12", size = 444513, upload-time = "2025-09-14T22:18:20.61Z" }, { url = "https://files.pythonhosted.org/packages/3a/43/241f9615bcf8ba8903b3f0432da069e857fc4fd1783bd26183db53c4804b/zstandard-0.25.0-cp314-cp314-win_amd64.whl", hash = "sha256:c19bcdd826e95671065f8692b5a4aa95c52dc7a02a4c5a0cac46deb879a017a2", size = 516118, upload-time = "2025-09-14T22:18:17.849Z" }, { url = "https://files.pythonhosted.org/packages/f0/ef/da163ce2450ed4febf6467d77ccb4cd52c4c30ab45624bad26ca0a27260c/zstandard-0.25.0-cp314-cp314-win_arm64.whl", hash = "sha256:d7541afd73985c630bafcd6338d2518ae96060075f9463d7dc14cfb33514383d", size = 476940, upload-time = "2025-09-14T22:18:19.088Z" }, -] +] \ No newline at end of file