transcribe

Installation
SKILL.md

Transcribe

Speech-to-text with two backends: MLX Whisper (local, Apple Silicon) and Groq API (cloud). Default mode auto uses local first, cloud as fallback.

Usage

bash scripts/transcribe.sh --input "/path/to/audio.m4a"

Options

Flag Description Default
--input FILE Audio/video file path (required)
--backend MODE auto, mlx, or groq auto
--model MODEL Whisper model name Per backend
--language LANG Language code (zh, en, ja, etc.) zh
Related skills
Installs
1
Repository
kcchien/skills
GitHub Stars
13
First Seen
Apr 16, 2026