Transcribe audio and video files using ElevenLabs Scribe, OpenAI Whisper, or Google Gemini. Supports automatic chunking for large files, speaker diarization, timestamps, and multiple output formats (text, SRT, VTT, JSON).
This skill provides high-quality speech-to-text transcription using multiple AI providers. It automatically handles large files through compression and chunking.
bun run src/index.ts transcribe \
--provider openai \
--input ./recording.mp3
bun run src/index.ts transcribe \
--provider elevenlabs \
--input ./meeting.mp3 \
--diarize \
--timestamps \
--format srt
bun run src/index.ts transcribe \
--provider gemini \
--input ./video.mp4 \
--format vtt \
--output ./captions.vtt
bun run src/index.ts providers
| Format | Extension | Description |
|---|---|---|
| text | .txt | Plain text transcript |
| srt | .srt | SubRip subtitle format |
| vtt | .vtt | WebVTT subtitle format |
| json | .json | Full structured data with metadata |
The skill automatically handles files larger than provider limits:
# ElevenLabs
export ELEVENLABS_API_KEY=your_key
# OpenAI
export OPENAI_API_KEY=your_key
# Google Gemini
export GOOGLE_API_KEY=your_key
For chunking support (OpenAI with large files):
ffmpeg - Audio processingffprobe - Duration detectionInstall on macOS:
brew install ffmpeg