v0.2.19

semaphoreui/semaphorev0.2.19Feb 23, 2026by QuentinFuxa

AI Summary

The release adds a new Voxtral backend optimized for Apple Silicon and Linux/GPU, alongside a comprehensive benchmarking suite for evaluating ASR performance.

Key Highlights

New Voxtral backend available in `voxtral-mlx` (Apple Silicon) and `voxtral (HF)` (Linux/GPU) variants.
Introduction of an offline benchmark harness for measuring WER, RTF, and timestamp accuracy.
New benchmark report in BENCHMARK.md with tables, charts, and recommendations.

New Features

Voxtral Native MLX Backend
Voxtral HuggingFace Backend
Offline Benchmark Harness

Full Release Notes

## Voxtral backend & benchmarks

### New: Voxtral backend

- **voxtral-mlx**: Native MLX backend for Apple Silicon. Runs at 0.18-0.32x real-time, handles 100+ languages with automatic language detection. No extra dependencies needed on macOS.
- **voxtral (HF)**: HuggingFace transformers backend for Linux/GPU. Requires `pip install transformers torch`.

### Benchmarks

New offline benchmark harness (`test_backend_offline.py --benchmark`) that runs all installed backends and computes WER, RTF, and timestamp accuracy against ground truth transcripts. Results exportable as JSON.

Full benchmark report in [BENCHMARK.md](https://github.com/QuentinFuxa/WhisperLiveKit/blob/main/BENCHMARK.md) with tables, charts, and recommendations for every backend/policy/model combination.

### Bug fixes

- Fixed silence double-counting in the audio processor
- Fixed median calculation for even-length lists in timestamp accuracy
- Fixed RTF inflation in metrics collector (was using wall-clock time instead of ASR processing time)