Transcribe

AI transcription skill for Claude Code - Transform audio/video recordings into transcripts with speaker diarization, AI-powered summaries, and visual infographics.

This skill provides a complete pipeline for processing recordings:

• Transcription - Convert audio/video to VTT format with speaker identification (OpenAI Whisper)
• Summarization - Generate structured markdown summaries (OpenAI GPT-5.1)
• Infographics - Create visual summaries from text (Google Gemini)
• All-in-one - Process video → transcript → summary → infographic in one command

See skills/transcribe/SKILL.md for complete usage guide.

Use in Claude Code

This is a Claude Code skill. Install it from the marketplace:

/plugin marketplace add krasnoperov/claude-plugins
/plugin install transcribe@krasnoperov-plugins

Once installed, use the /transcribe skill in your conversations:

/transcribe transcribe meeting.mp4 to VTT with speaker diarization
/transcribe summarize this transcript into key points
/transcribe create an infographic from this summary

Command Line Usage

You can also use this package directly via npx:

export OPENAI_API_KEY="your-openai-key"
export GOOGLE_AI_STUDIO_KEY="your-google-key"

# Transcribe audio/video
npx -y @krasnoperov/transcribe@latest transcribe meeting.mp4 -o transcript.vtt

# Generate summary
npx -y @krasnoperov/transcribe@latest summarize transcript.vtt -o summary.md

# Create infographic
npx -y @krasnoperov/transcribe@latest infographic summary.md -o visual.png

# All-in-one pipeline
npx -y @krasnoperov/transcribe@latest process recording.mp4 --output-dir ./output

Get your API keys:

• OpenAI
• Google AI Studio

Core Operations

transcribe <input>           Audio/Video → VTT transcript with speakers
summarize <input>            Text/VTT → Markdown summary
infographic <input>          Text → Visual infographic image
process <input>              All-in-one: video → transcript → summary → infographic

These operations can be used individually or chained together.

Examples

See skills/transcribe/examples/ directory:

1. 01-basic-workflow.sh - Step-by-step transcription pipeline
2. 02-all-in-one.sh - Single command processing

Transcription with Speaker Diarization

npx -y @krasnoperov/transcribe@latest transcribe podcast.mp3 \
  --language es \
  --model gpt-4o-transcribe-diarize \
  -o podcast.vtt

Output (VTT with speaker tags):

WEBVTT

00:00:00.000 --> 00:00:02.450
<v A>Welcome to the podcast...

00:00:02.850 --> 00:00:08.200
<v B>Thanks for having me...

Custom Summarization

npx -y @krasnoperov/transcribe@latest summarize transcript.vtt \
  --prompt "Focus on action items and decisions" \
  -o summary.md

Styled Infographic

npx -y @krasnoperov/transcribe@latest infographic summary.md \
  --style "modern minimal corporate" \
  -o infographic.png

Options

Transcribe

--model <model>          gpt-4o-transcribe-diarize (default), gpt-4o-transcribe, whisper-1
--language <lang>        Language code (en, es, ru, de, etc.)
-o, --output <file>      Output VTT file

Summarize

--prompt <text>          Custom summarization instructions
-o, --output <file>      Output markdown file

Infographic

--style <text>           Style instructions for visual
--reference <image>      Reference image for style
-o, --output <file>      Output image file

Process (All-in-one)

--output-dir <dir>       Output directory for all files
--language <lang>        Language for transcription
--model <model>          Transcription model
--style <text>           Style for infographic

Requirements

• Node.js >= 18.0.0
• ffmpeg (for audio extraction)

# macOS
brew install ffmpeg

# Ubuntu/Debian
sudo apt install ffmpeg

Development

npm run build      # Build TypeScript
npm run typecheck  # Type checking
npm run test       # Run tests
npm run dev        # Dev mode with type stripping

License

MIT License - Copyright (c) 2025 Aleksei Krasnoperov

See LICENSE file for details.