ravindranathpathi

Watch a video (URL or local path). Picks frames at scene changes, OCRs each frame so on-screen text is read as text not pixels, pulls a timestamped transcript (captions or Whisper, auto-chunked for long audio), and hands it all to Claude.