huanglizhuo/QwenASR

Qwen3-ASR speech recognition on Apple Silicon via MLX

Rust implementation of Qwen3-ASR automatic speech recognition

C inference for Qwen3-ASR 0.6b and 1.7b transcriptions models

Optimized Whisper models for streaming and on-device use

Phonon: open speech recognition models (Phonon-2, Phonon-1) — CLI, CPU and CUDA images

Faster Whisper transcription with CTranslate2

A lightweight ComfyUI custom node pack for Qwen3-ASR, providing simple speech‑to‑text workflows with local model caching and optional timestamp output. Supports Qwen/Qwen3‑ASR‑1.7B and 0.6B, with HuggingFace/ModelScope download options and clean integration for ComfyUI pipelines.

Agent-first Rust ASR orchestration stack: Bayesian backend routing across whisper.cpp/insanely-fast-whisper/whisper-diarization, real-time NDJSON streaming, SQLite persistence, TTY audio transport, conformance harness. 107K lines, 2000+ tests, zero unsafe code.

An extremely fast implementation of whisper optimized for Apple Silicon using MLX.