Qwen ASR Skill

Speech-to-text using Qwen3-ASR model, running locally on Apple Silicon via a FastAPI service.

Overview

• Model: mlx-community/Qwen3-ASR-0.6B-4bit (4-bit quantized, ~400MB)
• Runtime: MLX (Apple Silicon GPU acceleration via MPS)
• Service: FastAPI on http://localhost:8100
• Languages: Chinese, English, Japanese, Korean, and more

First-time Deployment

Prerequisites