AgentSkillsCN

mlx-stt

在本地使用 MLX（Apple Silicon）和 GLM-ASR-Nano-2512 实现语音转文本。

SKILL.md
--- frontmatter
name: mlx-stt
description: Speech-To-Text with MLX (Apple Silicon) and GLM-ASR-Nano-2512 locally.
metadata: {"openclaw":{"always":true,"emoji":"🦞","homepage":"https://github.com/guoqiao/skills/blob/main/mlx-stt/mlx-stt/SKILL.md","os":["darwin"],"tags":["latest","asr","stt","speech-to-text","audio","glm","glm-asr","glm-asr-nano-2512","glm-asr-nano-2512-8bit","macOS","MacBook","Mac mini","Apple Silicon","mlx","mlx-audio"],"requires":{"bins":["brew"]}}}

MLX STT

Speech-To-Text/ASR/Transcribe with MLX (Apple Silicon) and GLM-ASR-Nano-2512 locally.

Free and Accurate. No api key required. No server required.

Requirements

  • •mlx: macOS with Apple Silicon
  • •brew: used to install deps if not available

Installation

bash
bash ${baseDir}/install.sh

This script will use brew to install these cli tools if not available:

  • •ffmpeg: convert audio format when needed
  • •uv: install python package and run python script
  • •mlx_audio: do the real job

Usage

To transcribe an audio file, run the mlx-stt.py script:

bash
uv run  ${baseDir}/mlx-stt.py <audio_file_path>
  • •When first run, it will download model from Hugging Face, default: mlx-community/GLM-ASR-Nano-2512-8bit, 2.5GB ish.
  • •The transcript result will be printed to stdout.