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

Fix Soniox processing metrics to measure token-to-transcript time

Browse Source 
Move start_processing_metrics from run_stt (called per audio chunk,
producing noisy 0ms logs) to _receive_messages when the first final
token arrives for a new utterance. The existing stop_processing_metrics
in send_endpoint_transcript completes the pair, giving a meaningful
measurement of time from first recognition to finalized transcript.
This commit is contained in:
Mark Backman
2026-02-24 13:09:29 -05:00
parent 6f7664846c
commit 23ad181515
1 changed files with 2 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

4
src/pipecat/services/soniox/stt.py
View File

@@ -301,10 +301,8 @@ class SonioxSTTService(WebsocketSTTService):
Yields: Yields:
Frame: None (transcription results come via WebSocket callbacks). Frame: None (transcription results come via WebSocket callbacks).
""" """
await self.start_processing_metrics()
if self._websocket and self._websocket.state is State.OPEN: if self._websocket and self._websocket.state is State.OPEN:
await self._websocket.send(audio) await self._websocket.send(audio)
await self.stop_processing_metrics()
yield None yield None
@@ -485,6 +483,8 @@ class SonioxSTTService(WebsocketSTTService):
# the rest will be sent as interim tokens (even final tokens). # the rest will be sent as interim tokens (even final tokens).
await send_endpoint_transcript() await send_endpoint_transcript()
else: else:
if not self._final_transcription_buffer:
await self.start_processing_metrics()
self._final_transcription_buffer.append(token) self._final_transcription_buffer.append(token)
else: else:
non_final_transcription.append(token) non_final_transcription.append(token)