transcribe-tool

$npx mdskill add xuiltul/animaworks/transcribe-tool

Transcribe audio files to text instantly using Whisper.

  • Converts meetings and podcasts into readable text.
  • Depends on faster-whisper and optional GPU acceleration.
  • Selects language and model based on user input.
  • Delivers raw text or LLM-polished transcripts.
SKILL.md
.github/skills/transcribe-toolView on GitHub ↗
---
name: transcribe-tool
description: >-
  Audio transcription tool. Converts audio files to text with Whisper and optional LLM post-processing.
  Use when: transcribing meetings, podcasts, or extracting text from recorded audio files.
tags: [audio, transcription, whisper, external]
---

# Transcribe Tool

External tool for speech-to-text using Whisper (faster-whisper).

## Invocation via Bash

Use **Bash** with `animaworks-tool transcribe <subcommand> [args]`:

### audio — Transcribe audio file
```bash
animaworks-tool transcribe transcribe audio_file.wav [-l ja] [-m large-v3-turbo]
```

## Parameters

| Parameter | Type | Default | Description |
|-----------|------|---------|--------------|
| audio_path | string | (required) | Path to audio file |
| language | string | null | Language code (ja, en, etc.). null for auto-detect |
| model | string | "large-v3-turbo" | Whisper model name |
| raw | boolean | false | If true, skip LLM post-processing |

## CLI Usage (S/C/D/G-mode)

```bash
animaworks-tool transcribe transcribe audio_file.wav [-l ja] [-m large-v3-turbo]
```

## Notes

- faster-whisper must be installed
- CUDA-compatible ctranslate2 required for GPU acceleration
- Model is auto-downloaded on first run
More from xuiltul/animaworks