# antirez/voxtral.c

Pure C inference of Mistral Voxtral Realtime 4B speech to text model

Repository: https://github.com/antirez/voxtral.c
Canonical: https://ross.abutalabs.com/products/voxtralc
Language: C
License: MIT
License Family: permissive
Last push: 2026-02-15T18:58:52+00:00

## Health v2 (maintenance only)
Score: 45/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 67, release rhythm 35, longevity 14
- inputs: {"age_days": 209, "days_push": 199, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1738, forks 121 (observed 2026-08-28T04:05:29.945151+00:00)

## What it is
A pure C, zero-dependency inference implementation of Mistral's Voxtral Realtime 4B speech-to-text model, with a streaming C API and CLI for transcribing files, stdin pipes, or live microphone input. It also ships a self-contained Python reference implementation for understanding the model pipeline.

## Use cases
- transcribe audio files to text locally
- real-time speech-to-text from microphone
- stream transcription of any audio format via ffmpeg
- embed speech recognition in C applications
- understand how the Voxtral model inference pipeline works
- run speech-to-text without Python or CUDA at inference time

## When to choose
- you need dependency-free, self-contained speech-to-text inference in C
- you want streaming, low-latency transcription on Apple Silicon via MPS
- you want a readable reference implementation of the Voxtral model

## When to avoid
- you need production-grade, battle-tested transcription accuracy
- you need GPU inference on CUDA or Linux-native fast backends
- you need to stress-test very long transcriptions with the KV cache

## Facets
- artifact type: library
- maturity: experimental
- function: speech-recognition, llm-inference, audio-processing, machine-learning
- domain: speech-processing, machine-learning, developer-tools
- platform: cli, cpp
- tags: speech-to-text, transcription, voxtral, pure-c, streaming-inference, mistral, self-contained, no-dependencies, natural-language-processing, macos, linux

## Member repositories
- antirez/voxtral.c (main) score 45

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:29.945151+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T03:30:35.528036+00:00, confidence not recorded.
  - readme: https://github.com/antirez/voxtral.c (fetched 2026-08-28T04:05:29.945151+00:00, sha 22674e32d25a)
- Data as of 2026-08-30T08:39:29.467469+00:00.