voxtral.c: Pure C inference of Mistral Voxtral Realtime 4B speech to text model
Summary
Voxtral.c delivers a pure C inference pipeline for Mistral Voxtral Realtime 4B with zero external dependencies, enabling streaming transcription directly in C projects. It provides a streaming API, a Python reference, and benchmarks across MPS and BLAS backends, highlighting a lightweight option for real-time speech-to-text in AI-enabled workflows. Note: the project is early-stage and may require further testing before production use.