Starstruk Logo MarkStarstruk

v3.1

snakers4/silero-vadv3.1Dec 17, 2021by snakers4

AI Summary

This release introduces an ONNX version of the Silero VAD model, offering 2-3x faster performance on short audio chunks while maintaining model quality and a compact size.

Key Highlights

  • Introduction of ONNX model support
  • 2-3x faster performance on short audio chunks
  • Compact model size (~100k parameters)
  • Model quality parity with PyTorch release

Breaking Changes

  • ONNX model currently only supports 16kHz sample rate

New Features

  • ONNX model implementation
  • Compact model architecture
  • Performance optimization for short audio

Full Release Notes

We finally were able to port a model to ONNX:

- Compact model (~100k params);
- Both PyTorch and ONNX models are not quantized;
- Same quality model as the latest best PyTorch release;
- Only 16kHz available now (ONNX has some issues with if-statements and / or tracing vs scripting) with cryptic errors;
- In our tests, on short audios (chunks) ONNX is 2-3x faster than PyTorch (this is mitigated with larger batches or long audios);
- Audio examples and non-core models moved out of the repo to save space;