v6.2.2

Wei-Shaw/sub2apiv6.2.2Sep 17, 2026by adamnsandle

Upgrading?

AI Summary

This release adds an offline ONNX sequence model optimized to release the Global Interpreter Lock (GIL), allowing for more efficient processing, specifically supporting 16 kHz audio.

Key Highlights

  • New `silero_vad_16k_sequence.onnx` model added.
  • New function `get_speech_timestamps_sequence` for sequence processing.
  • Optimized for 16 kHz sampling rate.
  • Offline processing with GIL release.

New Features

  • New `get_speech_timestamps_sequence` function.
  • New `silero_vad_16k_sequence.onnx` model.

Full Release Notes

Added offline ONNX sequence to release GIL- #795
- 16 kHz only for now
- separate `silero_vad_16k_sequence.onnx` model

```python3
from silero_vad import load_silero_vad, get_speech_timestamps_sequence, read_audio

model = load_silero_vad(sequence=True)          # loads silero_vad_16k_sequence.onnx
audio = read_audio("audio.wav", sampling_rate=16000)
ts = get_speech_timestamps_sequence(audio, model, return_seconds=True)
```

Related Repositories