What problem does it solve? It automates downloading videos from Douyin, Kuaishou, Bilibili, or direct links and turning them into transcripts, summaries, timelines, and frame-level visual evidence without manual transcription or scrubbing through footage. ## Core Features & Use Cases - Video Downloading: Resolve share links and short links from Douyin, Kuaishou, Bilibili, or direct video URLs and save local MP4 files with structured metadata. - Transcription & Summaries: Convert video audio with PyAV and upload to Lark Minutes to produce transcripts, subtitles, summaries, keywords, and timelines, with a fast Douyin text-only path via web.fetch. - Visual Understanding: Match transcript keywords to time windows and extract frames with PyAV to answer whether objects, people, or actions appear in a video. - Use Case: Given a Bilibili share link, download the MP4, generate a timestamped transcript via Lark Minutes, then extract frames around the moment a specific product appears to verify its presence. ## Quick Start Ask the assistant to extract the transcript and summary from a Douyin, Kuaishou, or Bilibili video link you provide.
