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

Add debug logging to track transcription generation

Browse Source 
- Add logging when new text is set and transcription flag is reset
- Add logging when transcription is generated vs skipped
- Add reset_word_timestamps() call to ensure clean state for new text
- This will help identify why transcriptions are still being repeated
This commit is contained in:
James Hush
2025-09-17 14:54:56 +08:00
parent 67cf4b400a
commit 77d8abcce5
1 changed files with 11 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
src/pipecat/services/asyncai/tts.py
View File

@@ -292,12 +292,20 @@ class AsyncAITTSService(WebsocketWordTTSService):
# Start word timestamps and add the entire text as one "word" # Start word timestamps and add the entire text as one "word"
# This generates transcription frames for the bot - but only once per text # This generates transcription frames for the bot - but only once per text
if not self._transcription_generated and self._current_text.strip(): if not self._transcription_generated and self._current_text.strip():
logger.debug(
f"{self}: Generating transcription for text: '{self._current_text}'"
)
self.start_word_timestamps() self.start_word_timestamps()
# Add the entire text as a single timestamp at time 0 # Add the entire text as a single timestamp at time 0
await self.add_word_timestamps([(self._current_text, 0.0)]) await self.add_word_timestamps([(self._current_text, 0.0)])
# Add stop markers to end the word timestamps # Add stop markers to end the word timestamps
await self.add_word_timestamps([("TTSStoppedFrame", 0), ("Reset", 0)]) await self.add_word_timestamps([("TTSStoppedFrame", 0), ("Reset", 0)])
self._transcription_generated = True self._transcription_generated = True
logger.debug(f"{self}: Transcription flag set to True")
else:
logger.debug(
f"{self}: Skipping transcription generation - already generated: {self._transcription_generated}, text: '{self._current_text}'"
)
frame = TTSAudioRawFrame( frame = TTSAudioRawFrame(
audio=base64.b64decode(msg["audio"]), audio=base64.b64decode(msg["audio"]),
@@ -352,6 +360,9 @@ class AsyncAITTSService(WebsocketWordTTSService):
# Store the current text for generating timestamps # Store the current text for generating timestamps
self._current_text = text self._current_text = text
self._transcription_generated = False # Reset for new text self._transcription_generated = False # Reset for new text
# Reset word timestamps for new text to ensure clean state
self.reset_word_timestamps()
logger.debug(f"{self}: New text set: '{text}', transcription flag reset to False")
msg = self._build_msg(text=text, force=True) msg = self._build_msg(text=text, force=True)
try: try: