dv-pipecat-ai 0.0.75.dev887__py3-none-any.whl → 0.0.82.dev19__py3-none-any.whl

pipecat/services/cartesia/tts.py

@@ -10,7 +10,7 @@ import base64
 import json
 import uuid
 import warnings
-from typing import AsyncGenerator, List, Optional, Union
+from typing import AsyncGenerator, List, Literal, Optional, Union
 
 from loguru import logger
 from pydantic import BaseModel, Field
@@ -102,7 +102,7 @@ class CartesiaTTSService(AudioContextWordTTSService):
 
         Parameters:
             language: Language to use for synthesis.
-            speed: Voice speed control (string or float).
+            speed: Voice speed control.
             emotion: List of emotion controls.
 
                 .. deprecated:: 0.0.68
@@ -110,7 +110,7 @@ class CartesiaTTSService(AudioContextWordTTSService):
         """
 
         language: Optional[Language] = Language.EN
-        speed: Optional[Union[str, float]] = ""
+        speed: Optional[Literal["slow", "normal", "fast"]] = None
         emotion: Optional[List[str]] = []
 
     def __init__(
@@ -272,11 +272,13 @@ class CartesiaTTSService(AudioContextWordTTSService):
         voice_config["id"] = self._voice_id
 
         if self._settings["emotion"]:
-            warnings.warn(
-                "The 'emotion' parameter in __experimental_controls is deprecated and will be removed in a future version.",
-                DeprecationWarning,
-                stacklevel=2,
-            )
+            with warnings.catch_warnings():
+                warnings.simplefilter("always")
+                warnings.warn(
+                    "The 'emotion' parameter in __experimental_controls is deprecated and will be removed in a future version.",
+                    DeprecationWarning,
+                    stacklevel=2,
+                )
             voice_config["__experimental_controls"] = {}
             if self._settings["emotion"]:
                 voice_config["__experimental_controls"]["emotion"] = self._settings["emotion"]
@@ -387,7 +389,7 @@ class CartesiaTTSService(AudioContextWordTTSService):
         await self._websocket.send(msg)
         self._context_id = None
 
-    async def _receive_messages(self):
+    async def _process_messages(self):
         async for message in self._get_websocket():
             msg = json.loads(message)
             if not msg or not self.audio_context_available(msg["context_id"]):
@@ -421,6 +423,14 @@ class CartesiaTTSService(AudioContextWordTTSService):
             else:
                 logger.error(f"{self} error, unknown message type: {msg}")
 
+    async def _receive_messages(self):
+        while True:
+            await self._process_messages()
+            # Cartesia times out after 5 minutes of innactivity (no keepalive
+            # mechanism is available). So, we try to reconnect.
+            logger.debug(f"{self} Cartesia connection was disconnected (timeout?), reconnecting")
+            await self._connect_websocket()
+
     @traced_tts
     async def run_tts(self, text: str) -> AsyncGenerator[Frame, None]:
         """Generate speech from text using Cartesia's streaming API.
@@ -472,7 +482,7 @@ class CartesiaHttpTTSService(TTSService):
 
         Parameters:
             language: Language to use for synthesis.
-            speed: Voice speed control (string or float).
+            speed: Voice speed control.
             emotion: List of emotion controls.
 
                 .. deprecated:: 0.0.68
@@ -480,7 +490,7 @@ class CartesiaHttpTTSService(TTSService):
         """
 
         language: Optional[Language] = Language.EN
-        speed: Optional[Union[str, float]] = ""
+        speed: Optional[Literal["slow", "normal", "fast"]] = None
         emotion: Optional[List[str]] = Field(default_factory=list)
 
     def __init__(
@@ -600,11 +610,13 @@ class CartesiaHttpTTSService(TTSService):
             voice_config = {"mode": "id", "id": self._voice_id}
 
             if self._settings["emotion"]:
-                warnings.warn(
-                    "The 'emotion' parameter in voice.__experimental_controls is deprecated and will be removed in a future version.",
-                    DeprecationWarning,
-                    stacklevel=2,
-                )
+                with warnings.catch_warnings():
+                    warnings.simplefilter("always")
+                    warnings.warn(
+                        "The 'emotion' parameter in voice.__experimental_controls is deprecated and will be removed in a future version.",
+                        DeprecationWarning,
+                        stacklevel=2,
+                    )
                 voice_config["__experimental_controls"] = {"emotion": self._settings["emotion"]}
 
             await self.start_ttfb_metrics()

pipecat/services/cerebras/llm.py

@@ -9,9 +9,8 @@
 from typing import List
 
 from loguru import logger
-from openai.types.chat import ChatCompletionMessageParam
 
-from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
+from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
 from pipecat.services.openai.llm import OpenAILLMService
 
 
@@ -27,7 +26,7 @@ class CerebrasLLMService(OpenAILLMService):
         *,
         api_key: str,
         base_url: str = "https://api.cerebras.ai/v1",
-        model: str = "llama-3.3-70b",
+        model: str = "gpt-oss-120b",
         **kwargs,
     ):
         """Initialize the Cerebras LLM service.
@@ -35,7 +34,7 @@ class CerebrasLLMService(OpenAILLMService):
         Args:
             api_key: The API key for accessing Cerebras's API.
             base_url: The base URL for Cerebras API. Defaults to "https://api.cerebras.ai/v1".
-            model: The model identifier to use. Defaults to "llama-3.3-70b".
+            model: The model identifier to use. Defaults to "gpt-oss-120b".
             **kwargs: Additional keyword arguments passed to OpenAILLMService.
         """
         super().__init__(api_key=api_key, base_url=base_url, model=model, **kwargs)
@@ -54,25 +53,31 @@ class CerebrasLLMService(OpenAILLMService):
         logger.debug(f"Creating Cerebras client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
 
-    def build_chat_completion_params(
-        self, context: OpenAILLMContext, messages: List[ChatCompletionMessageParam]
-    ) -> dict:
+    def build_chat_completion_params(self, params_from_context: OpenAILLMInvocationParams) -> dict:
         """Build parameters for Cerebras chat completion request.
 
         Cerebras supports a subset of OpenAI parameters, focusing on core
         completion settings without advanced features like frequency/presence penalties.
+
+        Args:
+            params_from_context: Parameters, derived from the LLM context, to
+                use for the chat completion. Contains messages, tools, and tool
+                choice.
+
+        Returns:
+            Dictionary of parameters for the chat completion request.
         """
         params = {
             "model": self.model_name,
             "stream": True,
-            "messages": messages,
-            "tools": context.tools,
-            "tool_choice": context.tool_choice,
             "seed": self._settings["seed"],
             "temperature": self._settings["temperature"],
             "top_p": self._settings["top_p"],
             "max_completion_tokens": self._settings["max_completion_tokens"],
         }
 
+        # Messages, tools, tool_choice
+        params.update(params_from_context)
+
         params.update(self._settings["extra"])
         return params

pipecat/services/deepgram/stt.py

@@ -276,6 +276,14 @@ class DeepgramSTTService(STTService):
     async def _disconnect(self):
         if self._connection.is_connected:
             self.logger.debug("Disconnecting from Deepgram")
+            # Deepgram swallows asyncio.CancelledError internally which prevents
+            # proper cancellation propagation. This issue was found with
+            # parallel pipelines where `CancelFrame` was not awaited for to
+            # finish in all branches and it was pushed downstream reaching the
+            # end of the pipeline, which caused `cleanup()` to be called while
+            # Deepgram disconnection was still finishing and therefore
+            # preventing the task cancellation that occurs during `cleanup()`.
+            # GH issue: https://github.com/deepgram/deepgram-python-sdk/issues/570
             await self._connection.finish()
 
     async def start_metrics(self):

pipecat/services/deepseek/llm.py

@@ -9,9 +9,8 @@
 from typing import List
 
 from loguru import logger
-from openai.types.chat import ChatCompletionMessageParam
 
-from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
+from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
 from pipecat.services.openai.llm import OpenAILLMService
 
 
@@ -54,19 +53,22 @@ class DeepSeekLLMService(OpenAILLMService):
         logger.debug(f"Creating DeepSeek client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
 
-    def _build_chat_completion_params(
-        self, context: OpenAILLMContext, messages: List[ChatCompletionMessageParam]
-    ) -> dict:
+    def _build_chat_completion_params(self, params_from_context: OpenAILLMInvocationParams) -> dict:
         """Build parameters for DeepSeek chat completion request.
 
         DeepSeek doesn't support some OpenAI parameters like seed and max_completion_tokens.
+
+        Args:
+            params_from_context: Parameters, derived from the LLM context, to
+                use for the chat completion. Contains messages, tools, and tool
+                choice.
+
+        Returns:
+            Dictionary of parameters for the chat completion request.
         """
         params = {
             "model": self.model_name,
             "stream": True,
-            "messages": messages,
-            "tools": context.tools,
-            "tool_choice": context.tool_choice,
             "stream_options": {"include_usage": True},
             "frequency_penalty": self._settings["frequency_penalty"],
             "presence_penalty": self._settings["presence_penalty"],
@@ -75,5 +77,8 @@ class DeepSeekLLMService(OpenAILLMService):
             "max_tokens": self._settings["max_tokens"],
         }
 
+        # Messages, tools, tool_choice
+        params.update(params_from_context)
+
         params.update(self._settings["extra"])
         return params

pipecat/services/elevenlabs/__init__.py

@@ -9,5 +9,7 @@ import sys
 from pipecat.services import DeprecatedModuleProxy
 
 from .tts import *
+from .stt import *
+# Old
 
 sys.modules[__name__] = DeprecatedModuleProxy(globals(), "elevenlabs", "elevenlabs.tts")

pipecat/services/elevenlabs/stt.py

@@ -0,0 +1,351 @@
+#
+# Copyright (c) 2024–2025, Daily
+#
+# SPDX-License-Identifier: BSD 2-Clause License
+#
+
+"""ElevenLabs speech-to-text service implementation."""
+
+import asyncio
+from typing import AsyncGenerator, Optional
+
+from loguru import logger
+
+from pipecat.frames.frames import ErrorFrame, Frame, TranscriptionFrame
+from pipecat.services.stt_service import SegmentedSTTService
+from pipecat.transcriptions.language import Language
+from pipecat.utils.time import time_now_iso8601
+from pipecat.utils.tracing.service_decorators import traced_stt
+
+try:
+    from elevenlabs.client import ElevenLabs
+except ModuleNotFoundError as e:
+    logger.error(f"Exception: {e}")
+    logger.error("In order to use ElevenLabs, you need to `pip install pipecat-ai[elevenlabs]`.")
+    raise Exception(f"Missing module: {e}")
+
+
+def language_to_elevenlabs_language(language: Language) -> Optional[str]:
+    """Maps pipecat Language enum to ElevenLabs language codes.
+
+    Args:
+        language: A Language enum value representing the input language.
+
+    Returns:
+        str or None: The corresponding ElevenLabs language code, or None if not supported.
+    """
+    language_map = {
+        # English
+        Language.EN: "eng",
+        Language.EN_US: "eng",
+        Language.EN_GB: "eng",
+        Language.EN_AU: "eng",
+        Language.EN_CA: "eng",
+        Language.EN_IN: "eng",
+        Language.EN_IE: "eng",
+        Language.EN_NZ: "eng",
+        Language.EN_ZA: "eng",
+        Language.EN_SG: "eng",
+        Language.EN_HK: "eng",
+        Language.EN_PH: "eng",
+        Language.EN_KE: "eng",
+        Language.EN_NG: "eng",
+        Language.EN_TZ: "eng",
+        # Spanish
+        Language.ES: "spa",
+        Language.ES_ES: "spa",
+        Language.ES_MX: "spa",
+        Language.ES_AR: "spa",
+        Language.ES_CO: "spa",
+        Language.ES_CL: "spa",
+        Language.ES_VE: "spa",
+        Language.ES_PE: "spa",
+        Language.ES_EC: "spa",
+        Language.ES_GT: "spa",
+        Language.ES_CU: "spa",
+        Language.ES_BO: "spa",
+        Language.ES_DO: "spa",
+        Language.ES_HN: "spa",
+        Language.ES_PY: "spa",
+        Language.ES_SV: "spa",
+        Language.ES_NI: "spa",
+        Language.ES_CR: "spa",
+        Language.ES_PA: "spa",
+        Language.ES_UY: "spa",
+        Language.ES_PR: "spa",
+        Language.ES_US: "spa",
+        Language.ES_GQ: "spa",
+        # French
+        Language.FR: "fra",
+        Language.FR_FR: "fra",
+        Language.FR_CA: "fra",
+        Language.FR_BE: "fra",
+        Language.FR_CH: "fra",
+        # German
+        Language.DE: "deu",
+        Language.DE_DE: "deu",
+        Language.DE_AT: "deu",
+        Language.DE_CH: "deu",
+        # Italian
+        Language.IT: "ita",
+        Language.IT_IT: "ita",
+        # Portuguese
+        Language.PT: "por",
+        Language.PT_PT: "por",
+        Language.PT_BR: "por",
+        # Hindi
+        Language.HI: "hin",
+        Language.HI_IN: "hin",
+        # Arabic
+        Language.AR: "ara",
+        Language.AR_SA: "ara",
+        Language.AR_EG: "ara",
+        Language.AR_AE: "ara",
+        Language.AR_BH: "ara",
+        Language.AR_DZ: "ara",
+        Language.AR_IQ: "ara",
+        Language.AR_JO: "ara",
+        Language.AR_KW: "ara",
+        Language.AR_LB: "ara",
+        Language.AR_LY: "ara",
+        Language.AR_MA: "ara",
+        Language.AR_OM: "ara",
+        Language.AR_QA: "ara",
+        Language.AR_SY: "ara",
+        Language.AR_TN: "ara",
+        Language.AR_YE: "ara",
+        # Japanese
+        Language.JA: "jpn",
+        Language.JA_JP: "jpn",
+        # Korean
+        Language.KO: "kor",
+        Language.KO_KR: "kor",
+        # Chinese
+        Language.ZH: "cmn",
+        Language.ZH_CN: "cmn",
+        Language.ZH_TW: "cmn",
+        Language.ZH_HK: "cmn",
+        # Russian
+        Language.RU: "rus",
+        Language.RU_RU: "rus",
+        # Dutch
+        Language.NL: "nld",
+        Language.NL_NL: "nld",
+        Language.NL_BE: "nld",
+        # Polish
+        Language.PL: "pol",
+        Language.PL_PL: "pol",
+        # Turkish
+        Language.TR: "tur",
+        Language.TR_TR: "tur",
+        # Swedish
+        Language.SV: "swe",
+        Language.SV_SE: "swe",
+        # Norwegian
+        Language.NO: "nor",
+        Language.NB: "nor",
+        Language.NN: "nor",
+        # Danish
+        Language.DA: "dan",
+        Language.DA_DK: "dan",
+        # Finnish
+        Language.FI: "fin",
+        Language.FI_FI: "fin",
+        # Czech
+        Language.CS: "ces",
+        Language.CS_CZ: "ces",
+        # Hungarian
+        Language.HU: "hun",
+        Language.HU_HU: "hun",
+        # Greek
+        Language.EL: "ell",
+        Language.EL_GR: "ell",
+        # Hebrew
+        Language.HE: "heb",
+        Language.HE_IL: "heb",
+        # Thai
+        Language.TH: "tha",
+        Language.TH_TH: "tha",
+        # Vietnamese
+        Language.VI: "vie",
+        Language.VI_VN: "vie",
+        # Indonesian
+        Language.ID: "ind",
+        Language.ID_ID: "ind",
+        # Malay
+        Language.MS: "msa",
+        Language.MS_MY: "msa",
+        # Ukrainian
+        Language.UK: "ukr",
+        Language.UK_UA: "ukr",
+        # Bulgarian
+        Language.BG: "bul",
+        Language.BG_BG: "bul",
+        # Croatian
+        Language.HR: "hrv",
+        Language.HR_HR: "hrv",
+        # Slovak
+        Language.SK: "slk",
+        Language.SK_SK: "slk",
+        # Slovenian
+        Language.SL: "slv",
+        Language.SL_SI: "slv",
+        # Estonian
+        Language.ET: "est",
+        Language.ET_EE: "est",
+        # Latvian
+        Language.LV: "lav",
+        Language.LV_LV: "lav",
+        # Lithuanian
+        Language.LT: "lit",
+        Language.LT_LT: "lit",
+    }
+    return language_map.get(language)
+
+
+class ElevenlabsSTTService(SegmentedSTTService):
+    """ElevenLabs speech-to-text service using Scribe v1 model.
+
+    This service uses ElevenLabs' batch STT API to transcribe audio segments.
+    It extends SegmentedSTTService to handle VAD-based audio segmentation.
+
+    Args:
+        api_key: ElevenLabs API key for authentication.
+        model_id: Model to use for transcription (default: "scribe_v1").
+        language: Default language for transcription.
+        tag_audio_events: Whether to tag audio events like laughter (default: False).
+        diarize: Whether to enable speaker diarization (default: False).
+        **kwargs: Additional arguments passed to SegmentedSTTService.
+    """
+
+    def __init__(
+        self,
+        *,
+        api_key: str,
+        model_id: str = "scribe_v1",
+        language: Language = Language.EN,
+        tag_audio_events: bool = False,
+        sample_rate: Optional[int] = None,
+        diarize: bool = False,
+        **kwargs,
+    ):
+        super().__init__(**kwargs)
+
+        self._client = ElevenLabs(api_key=api_key)
+        self._model_id = model_id
+        self._tag_audio_events = tag_audio_events
+        self._diarize = diarize
+
+        self._settings = {
+            "language": language,
+            "model_id": self._model_id,
+            "tag_audio_events": self._tag_audio_events,
+            "diarize": self._diarize,
+        }
+        self.set_model_name(model_id)
+
+    def can_generate_metrics(self) -> bool:
+        """Check if this service can generate processing metrics.
+
+        Returns:
+            True, as ElevenLabs service supports metrics generation.
+        """
+        return True
+
+    def language_to_service_language(self, language: Language) -> Optional[str]:
+        """Convert from pipecat Language to ElevenLabs language code.
+
+        Args:
+            language: The Language enum value to convert.
+
+        Returns:
+            str or None: The corresponding ElevenLabs language code, or None if not supported.
+        """
+        return language_to_elevenlabs_language(language)
+
+    async def set_language(self, language: Language):
+        """Set the language for transcription.
+
+        Args:
+            language: The Language enum value to use for transcription.
+        """
+        self.logger.info(f"Switching STT language to: [{language}]")
+        self._settings["language"] = language
+
+    @traced_stt
+    async def _handle_transcription(
+        self, transcript: str, is_final: bool, language: Optional[Language] = None
+    ):
+        """Handle a transcription result with tracing."""
+        pass
+
+    async def run_stt(self, audio: bytes) -> AsyncGenerator[Frame, None]:
+        """Transcribe the provided audio using ElevenLabs STT.
+
+        Args:
+            audio: Audio data (WAV format) to transcribe.
+
+        Yields:
+            Frame: TranscriptionFrame containing the transcribed text or ErrorFrame on failure.
+        """
+        try:
+            await self.start_processing_metrics()
+            await self.start_ttfb_metrics()
+
+            # Get language code for ElevenLabs API
+            language = self._settings["language"]
+            elevenlabs_lang = self.language_to_service_language(language)
+
+            # Prepare API parameters
+            params = {
+                "file": audio,
+                "model_id": self._model_id,
+                "tag_audio_events": self._tag_audio_events,
+                "diarize": self._diarize,
+            }
+
+            # Add language if specified
+            if elevenlabs_lang:
+                params["language_code"] = elevenlabs_lang
+
+            # Call ElevenLabs STT API in thread pool to avoid blocking
+            transcription = await asyncio.to_thread(self._client.speech_to_text.convert, **params)
+
+            await self.stop_ttfb_metrics()
+
+            # Process transcription result
+            if transcription and hasattr(transcription, "text") and transcription.text:
+                transcript_text = transcription.text.strip()
+
+                if transcript_text:
+                    # Determine language if available from response
+                    response_language = language
+                    if hasattr(transcription, "language_code") and transcription.language_code:
+                        # Try to map back from ElevenLabs language code to pipecat Language
+                        try:
+                            # This is a simplified mapping - you might want to create a reverse map
+                            response_language = language  # For now, keep the original
+                        except ValueError:
+                            self.logger.warning(
+                                f"Unknown language detected: {transcription.language_code}"
+                            )
+
+                    # Handle transcription with tracing
+                    await self._handle_transcription(transcript_text, True, response_language)
+
+                    self.logger.debug(f"ElevenLabs transcription: [{transcript_text}]")
+
+                    yield TranscriptionFrame(
+                        text=transcript_text,
+                        user_id="",
+                        timestamp=time_now_iso8601(),
+                        language=response_language,
+                        result=transcription,
+                    )
+
+            await self.stop_processing_metrics()
+
+        except Exception as e:
+            self.logger.error(f"ElevenLabs STT error: {e}")
+            await self.stop_all_metrics()
+            yield ErrorFrame(f"ElevenLabs STT error: {str(e)}")

pipecat/services/fireworks/llm.py

@@ -9,9 +9,8 @@
 from typing import List
 
 from loguru import logger
-from openai.types.chat import ChatCompletionMessageParam
 
-from pipecat.processors.aggregators.openai_llm_context import OpenAILLMContext
+from pipecat.adapters.services.open_ai_adapter import OpenAILLMInvocationParams
 from pipecat.services.openai.llm import OpenAILLMService
 
 
@@ -54,20 +53,23 @@ class FireworksLLMService(OpenAILLMService):
         logger.debug(f"Creating Fireworks client with api {base_url}")
         return super().create_client(api_key, base_url, **kwargs)
 
-    def build_chat_completion_params(
-        self, context: OpenAILLMContext, messages: List[ChatCompletionMessageParam]
-    ) -> dict:
+    def build_chat_completion_params(self, params_from_context: OpenAILLMInvocationParams) -> dict:
         """Build parameters for Fireworks chat completion request.
 
         Fireworks doesn't support some OpenAI parameters like seed, max_completion_tokens,
         and stream_options.
+
+        Args:
+            params_from_context: Parameters, derived from the LLM context, to
+                use for the chat completion. Contains messages, tools, and tool
+                choice.
+
+        Returns:
+            Dictionary of parameters for the chat completion request.
         """
         params = {
             "model": self.model_name,
             "stream": True,
-            "messages": messages,
-            "tools": context.tools,
-            "tool_choice": context.tool_choice,
             "frequency_penalty": self._settings["frequency_penalty"],
             "presence_penalty": self._settings["presence_penalty"],
             "temperature": self._settings["temperature"],
@@ -75,5 +77,8 @@ class FireworksLLMService(OpenAILLMService):
             "max_tokens": self._settings["max_tokens"],
         }
 
+        # Messages, tools, tool_choice
+        params.update(params_from_context)
+
         params.update(self._settings["extra"])
         return params

pipecat/services/fish/tts.py

@@ -120,12 +120,14 @@ class FishAudioTTSService(InterruptibleTTSService):
         if model:
             import warnings
 
-            warnings.warn(
-                "Parameter 'model' is deprecated and will be removed in a future version. "
-                "Use 'reference_id' instead.",
-                DeprecationWarning,
-                stacklevel=2,
-            )
+            with warnings.catch_warnings():
+                warnings.simplefilter("always")
+                warnings.warn(
+                    "Parameter 'model' is deprecated and will be removed in a future version. "
+                    "Use 'reference_id' instead.",
+                    DeprecationWarning,
+                    stacklevel=2,
+                )
             reference_id = model
 
         self._api_key = api_key

pipecat/services/gemini_multimodal_live/gemini.py

@@ -33,6 +33,7 @@ from pipecat.frames.frames import (
     InputAudioRawFrame,
     InputImageRawFrame,
     InputTextRawFrame,
+    LLMContextFrame,
     LLMFullResponseEndFrame,
     LLMFullResponseStartFrame,
     LLMMessagesAppendFrame,
@@ -738,6 +739,10 @@ class GeminiMultimodalLiveLLMService(LLMService):
                 # Support just one tool call per context frame for now
                 tool_result_message = context.messages[-1]
                 await self._tool_result(tool_result_message)
+        elif isinstance(frame, LLMContextFrame):
+            raise NotImplementedError(
+                "Universal LLMContext is not yet supported for Gemini Multimodal Live."
+            )
         elif isinstance(frame, InputTextRawFrame):
             await self._send_user_text(frame.text)
             await self.push_frame(frame, direction)