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

Call start_word_timestamps() when the first audio chunk arrives

Browse Source
This commit is contained in:
yukiobata1
2026-01-07 13:10:41 +09:00
parent 137bbb3d2c
commit f62c262f23
1 changed files with 7 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/pipecat/services/azure/tts.py
View File

@@ -441,9 +441,9 @@ class AzureTTSService(WordTTSService, AzureBaseTTSService):
try: try:
if not self._started: if not self._started:
await self.start_ttfb_metrics() await self.start_ttfb_metrics()
await self.start_word_timestamps()
yield TTSStartedFrame() yield TTSStartedFrame()
self._started = True self._started = True
self._first_chunk = True
self._cumulative_audio_offset = 0.0 self._cumulative_audio_offset = 0.0
ssml = self._construct_ssml(text) ssml = self._construct_ssml(text)
@@ -457,6 +457,12 @@ class AzureTTSService(WordTTSService, AzureBaseTTSService):
break break
await self.stop_ttfb_metrics() await self.stop_ttfb_metrics()
# Start word timestamps when first chunk arrives
if self._first_chunk:
await self.start_word_timestamps()
self._first_chunk = False
frame = TTSAudioRawFrame( frame = TTSAudioRawFrame(
audio=chunk, audio=chunk,
sample_rate=self.sample_rate, sample_rate=self.sample_rate,