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

SegmentedSTTService: use VAD user started/stopped speaking frames

Browse Source
This commit is contained in:
Aleix Conchillo Flaqué
2025-12-14 12:27:53 -08:00
parent 49ebe34599
commit 222ccbb471
1 changed files with 6 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files

17
src/pipecat/services/stt_service.py
View File

@@ -20,8 +20,8 @@ from pipecat.frames.frames import (
StartFrame, StartFrame,
STTMuteFrame, STTMuteFrame,
STTUpdateSettingsFrame, STTUpdateSettingsFrame,
UserStartedSpeakingFrame, VADUserStartedSpeakingFrame,
UserStoppedSpeakingFrame, VADUserStoppedSpeakingFrame,
) )
from pipecat.processors.frame_processor import FrameDirection from pipecat.processors.frame_processor import FrameDirection
from pipecat.services.ai_service import AIService from pipecat.services.ai_service import AIService
@@ -252,20 +252,15 @@ class SegmentedSTTService(STTService):
"""Process frames, handling VAD events and audio segmentation.""" """Process frames, handling VAD events and audio segmentation."""
await super().process_frame(frame, direction) await super().process_frame(frame, direction)
if isinstance(frame, UserStartedSpeakingFrame): if isinstance(frame, VADUserStartedSpeakingFrame):
await self._handle_user_started_speaking(frame) await self._handle_user_started_speaking(frame)
elif isinstance(frame, UserStoppedSpeakingFrame): elif isinstance(frame, VADUserStoppedSpeakingFrame):
await self._handle_user_stopped_speaking(frame) await self._handle_user_stopped_speaking(frame)
async def _handle_user_started_speaking(self, frame: UserStartedSpeakingFrame): async def _handle_user_started_speaking(self, frame: VADUserStartedSpeakingFrame):
if frame.emulated:
return
self._user_speaking = True self._user_speaking = True
async def _handle_user_stopped_speaking(self, frame: UserStoppedSpeakingFrame): async def _handle_user_stopped_speaking(self, frame: VADUserStoppedSpeakingFrame):
if frame.emulated:
return
self._user_speaking = False self._user_speaking = False
content = io.BytesIO() content = io.BytesIO()