Files
pipecat/examples/foundational/28-transcription-processor.py
James Hush 0a163201ea feat: Add sentence aggregation and Whisker debugger to transcript processor
- Enhance TranscriptHandler to aggregate transcript fragments into complete sentences using match_endofsentence()
- Add Whisker debugger integration for real-time pipeline visualization
- Implement sentence buffering for both user and assistant messages
- Add finalize_partial_sentences() method to handle incomplete sentences on disconnect
- Improves transcript readability by reducing fragmented output

Changes:
- Import match_endofsentence utility for sentence boundary detection
- Add pipecat_whisker.WhiskerObserver for debugging capabilities
- Modify on_transcript_update() to accumulate and aggregate messages
- Create _save_sentence() helper method for complete sentence handling
- Update client disconnect handler to preserve partial sentences
2025-09-25 14:01:19 +08:00

9.9 KiB