A tool to download YouTube videos, transcribe them, analyze for viral moments, and generate short clips.

1. Clone the repo.
2. Create a virtual environment: python -m venv venv
3. Activate: source venv/bin/activate (Linux/Mac) or venv\Scripts\activate (Windows)
4. Install dependencies: pip install -r requirements.txt
5. Run: python run_pipeline.py "https://www.youtube.com/watch?v=..."

• --top-k: Number of clips (default 3)
• --max-duration: Max clip length (default 60s)
• --no-vertical: Keep original aspect ratio
• --no-face: Disable face tracking (always center crop)
• --cookies: Path to cookies file for yt-dlp
• --cookies-from-browser: Browser to extract cookies from

Clips are saved in data/clips/, subtitles in data/subtitles/, transcripts in data/transcripts/.

The data/ folder is gitignored as it contains generated files.

• Logs are written to data/logs/mitoclipper.log and stdout via core/logging_config.py.
• Metrics are exposed via:
  • /api/result (includes metrics with last run summary)
  • /api/metrics (direct metrics endpoint)
• Metrics file: data/transcripts/pipeline_metrics.json.