Skip to content

Return VAD sentence timestamps for SenseVoice - #3414

Merged
LauraGPT merged 1 commit into
mainfrom
codex/sensevoice-vad-sentence-info
Jul 24, 2026
Merged

Return VAD sentence timestamps for SenseVoice#3414
LauraGPT merged 1 commit into
mainfrom
codex/sensevoice-vad-sentence-info

Conversation

@LauraGPT

Copy link
Copy Markdown
Collaborator

Summary

  • return per-VAD sentence_info when sentence_timestamp=True and an ASR model such as SenseVoice has no token timestamps
  • preserve punctuation-based segmentation and timestamp-capable ASR behavior
  • include both text and sentence fields for downstream compatibility

Validation

  • 30 passed across VAD/punctuation, subtitle, CLI, and OpenAI server segment tests
  • real iic/SenseVoiceSmall CPU inference on asr_example.wav: VAD sentence 610-5530 ms, recognized expected Chinese text
  • py_compile and git diff --check pass
@gemini-code-assist

Copy link
Copy Markdown
Contributor

Caution

The consumer version of Gemini Code Assist on GitHub has been sunset. All code review activity has officially ceased.

@LauraGPT
LauraGPT merged commit 99b7030 into main Jul 24, 2026
@LauraGPT
LauraGPT deleted the codex/sensevoice-vad-sentence-info branch July 24, 2026 12:20
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

1 participant