What you get

Chapters follow the content, instead of a cut every five minutes.

Cutting YouTube Video by Meaning, Not by Clock

Added 2026-08-06

It downloads at up to 1080p and chapters the video from its captions by meaning rather than fixed duration, aiming at two-to-five-minute chapters; you pick the chapters before anything is cut or translated.

Design & VideoContent & Knowledge
Cutting YouTube Video by Meaning, Not by Clock preview

Each kept chapter produces four files: the raw cut, a subtitled version, a bilingual SRT, and social copy. Translation runs in batches of twenty by default, which the author says cuts API calls by 95 percent. One prerequisite comes first: on macOS, Homebrew's standard ffmpeg is built without libass, so subtitle burn-in always fails — install ffmpeg-full and verify with ffmpeg -filters; Ubuntu needs libass-dev. Note also that chaptering reads the video's own captions before the model analyses them, so a video without captions will not work.