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

Updated VAD

Browse Source
This commit is contained in:
Sam Sykes
2025-12-31 01:28:31 +00:00
parent 2e700c8576
commit f7c74cfa80
1 changed files with 17 additions and 9 deletions
Download Patch File Download Diff File Expand all files Collapse all files

26
src/pipecat/services/speechmatics/stt.py
View File

@@ -27,8 +27,8 @@ from pipecat.frames.frames import (
InterimTranscriptionFrame, InterimTranscriptionFrame,
StartFrame, StartFrame,
TranscriptionFrame, TranscriptionFrame,
UserStartedSpeakingFrame, VADUserStartedSpeakingFrame,
UserStoppedSpeakingFrame, VADUserStoppedSpeakingFrame,
) )
from pipecat.processors.frame_processor import FrameDirection from pipecat.processors.frame_processor import FrameDirection
from pipecat.services.stt_service import STTService from pipecat.services.stt_service import STTService
@@ -597,17 +597,21 @@ class SpeechmaticsSTTService(STTService):
The service will: The service will:
- Send a BotInterruptionFrame upstream to stop bot speech - Send a BotInterruptionFrame upstream to stop bot speech
- Send a UserStartedSpeakingFrame downstream to notify other components - Send a VADUserStartedSpeakingFrame downstream to notify other components
- Start metrics collection for measuring response times - Start metrics collection for measuring response times
Args: Args:
message: the message payload. message: the message payload.
""" """
logger.debug(f"{self} StartOfTurn received") logger.debug(f"{self} StartOfTurn received")
await self.broadcast_frame(UserStartedSpeakingFrame)
await self.push_interruption_task_frame_and_wait()
# await self.start_processing_metrics() # await self.start_processing_metrics()
# Emit VAD events if enabled
if self._enable_vad:
await self.push_interruption_task_frame_and_wait()
logger.debug(f"{self} sending VADUserStartedSpeakingFrame")
await self.broadcast_frame(VADUserStartedSpeakingFrame)
async def _handle_end_of_turn(self, message: dict[str, Any]) -> None: async def _handle_end_of_turn(self, message: dict[str, Any]) -> None:
"""Handle EndOfTurn events. """Handle EndOfTurn events.
@@ -618,14 +622,18 @@ class SpeechmaticsSTTService(STTService):
The service will: The service will:
- Stop processing metrics collection - Stop processing metrics collection
- Send a UserStoppedSpeakingFrame to signal turn completion - Send a VADUserStoppedSpeakingFrame to signal turn completion
Args: Args:
message: the message payload. message: the message payload.
""" """
logger.debug(f"{self} EndOfTurn received") logger.debug(f"{self} EndOfTurn received")
# await self.stop_processing_metrics() # await self.stop_processing_metrics()
await self.broadcast_frame(UserStoppedSpeakingFrame)
# Emit VAD events if enabled
if self._enable_vad:
logger.debug(f"{self} sending VADUserStoppedSpeakingFrame")
await self.broadcast_frame(VADUserStoppedSpeakingFrame)
async def _handle_speakers_result(self, message: dict[str, Any]) -> None: async def _handle_speakers_result(self, message: dict[str, Any]) -> None:
"""Handle SpeakersResult events. """Handle SpeakersResult events.
@@ -660,10 +668,10 @@ class SpeechmaticsSTTService(STTService):
self._bot_speaking = False self._bot_speaking = False
# Force finalization # Force finalization
if isinstance(frame, UserStoppedSpeakingFrame): if isinstance(frame, VADUserStoppedSpeakingFrame):
if self._enable_vad: if self._enable_vad:
logger.warning( logger.warning(
f"{self} UserStoppedSpeakingFrame received but internal VAD is being used" f"{self} VADUserStoppedSpeakingFrame received but internal VAD is being used"
) )
elif not self._enable_vad and self._client is not None: elif not self._enable_vad and self._client is not None:
self._client.finalize() self._client.finalize()