Starstruk Logo MarkStarstruk

0.2.1

QuentinFuxa/WhisperLiveKit0.2.1Jun 27, 2025by QuentinFuxa

AI Summary

Introduces a new SimulStreaming backend that offers up to 5x faster transcription speeds compared to standard methods.

Key Highlights

  • Added SimulStreaming backend for real-time transcription.
  • Achieved up to 5x speedup on the tiny model.
  • Dual-licensed under the MIT license.

New Features

  • SimulStreaming backend
  • Installation via pip extra
  • 5x speedup on tiny models

Full Release Notes

New [SimulStreaming](https://github.com/ufal/SimulStreaming) backend for transcription. Associated preprint: https://arxiv.org/abs/2506.17077 
- Up to 5 time faster on tiny model: https://github.com/QuentinFuxa/WhisperLiveKit/issues/134#issuecomment-3006435314
- Requires to install `pip install whisperlivekit[simulstreaming]`. Dual licensed: https://github.com/ufal/SimulStreaming?tab=readme-ov-file#-licence-and-contributions
- Use it with `--backend simulstreaming`
- SimulStreaming limitations for now:
  - No buffer preview is available 
  - Diarization maybe be less precise
  -  Punctuation can be less accurate
  - English-only model (tiny.en, base.en, medium.en) are not compatible for now