0.2.14

QuentinFuxa/WhisperLiveKit0.2.14Nov 15, 2025by QuentinFuxa

AI Summary

Major release with new CLI parameters, backend policy restructuring, and frontend improvements including translation buffer display.

Key Highlights

Translation buffer displayed in light grey in frontend
CLI can use 'wlk' command
--direct-english-translation flag (renamed from --task translate)
--backend-policy for streaming strategy selection
--backend as Whisper implementation selector
--model-path accepts Hugging Face repo names

New Features

Frontend translation buffer visualization
CLI 'wlk' command support
Backend policy restructuring
Hugging Face repo name support for model-path

Full Release Notes

## New Features

* When using [NLLW](https://github.com/QuentinFuxa/NoLanguageLeftWaiting) translation, the buffer is now displayed in light grey in the frontend.
* CLI / Parameters:
  * Can now be called with `wlk` instead of / in addition to `whisperlivekit-server`.
  * `--task translate` flag renamed to `--direct-english-translation`.
  * Backend parameters have been rethought:

    * `--backend-policy` sets the streaming strategy:
      `simulstreaming` uses AlignAtt SimulStreaming,
      `localagreement` uses the LocalAgreement policy.
    * `--backend` is now the Whisper implementation selector (SimulStreaming: encoder only; LocalAgreement: full encoder+decoder). `auto` selects the best config based on the system.
    * `--model-path` now accepts direct Hugging Face repo names (e.g. `qanastek/whisper-tiny-french-cased`, `KBLab/kb-whisper-tiny`, etc.).
  * For custom models, `scripts/determine_alignment_heads.py` determines the best alignment heads for SimulStreaming.

## Bug Fixes

* Fixed flickering between speakers when using diarization.
* Correct task resolution at the end of the stream.
* Fixed custom model handling for some models (e.g. #269).
*  **0.2.14.post1**: Improve silence handling, by producing remaining non-validated tokens when silence starts