A Codex skill turning Chinese livestreams, interviews, or long videos into publishable opinion clips: it transcribes with Doubao ASR, Whisper, or existing subtitles, selects highlights by semantic meaning rather than fixed-duration cuts, extracts a hook, removes filler and false starts, fixes ASR typos into captions, and batch-renders with FFmpeg.
STACK
FFmpegWhisperNode.js
“Open-source Codex skill for semantic Chinese video clipping with hooks, captions, cleanup, and FFmpeg rendering”
ai-videoasrcodex-skillffmpegvideo-clipping
About Editing
Tools that work on footage you already have: cutting, captions, B-roll, silence removal and recaps, often driving FFmpeg or an NLE.