v0.2.24

PaddlePaddle/Paddlev0.2.24Jul 11, 2026by QuentinFuxa

AI Summary

This release introduces the AlignAtt translation backend for streaming LLM translation, refreshes benchmark figures to correct previous inaccuracies, and improves error handling to fail loudly when the ASR backend produces no output.

Key Highlights

New AlignAtt translation backend for streaming LLM translation
Benchmark figures re-measured on H100
Fixed empty captions issue with mlx-whisper
Server now fails loudly on ASR backend errors

New Features

AlignAtt translation backend (`--translation-backend alignatt`)

Full Release Notes

Streaming LLM translation, refreshed benchmarks and loud failures.

## Added
- AlignAtt translation backend (`--translation-backend alignatt`): streaming LLM translation through an [Alignatt4LLM](https://github.com/QuentinFuxa/Alignatt4LLM) sidecar. The model drafts ahead over the unstable ASR tail and commits only target words whose attention lands on committed source words, so translations are append-only and released the instant the ASR commits. See [docs/translation-alignatt.md](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/docs/translation-alignatt.md).
- Benchmark figures re-measured on an H100 with the current backends (the README scatter plots dated from March and misrepresented the qwen3 backends). Raw results in `benchmarks/h100_scatter/`.

## Fixed
- mlx-whisper produced empty captions with torch >= 2.13 (#383, thanks @RobertBartelds-FleetEnergies).
- The server now fails loudly when the ASR backend produces nothing: warmup errors abort startup with the cause, and a watchdog logs an explicit error if audio flows but no text is ever produced.