Skip to main content
bthebigclass
All help topics

For creators

Auto-chapters from your transcript

Every recording auto-generates chapter markers from the transcript. Transition phrases like 'now let's talk about…' become seekable chapter points in the player.

Last updated May 21, 2026

How chapters are generated

When a recording's WebVTT transcript is available, we scan cues for transition markers — phrases like 'now let's talk about', 'moving on to', 'next up', 'finally'. Each match becomes a candidate chapter at the cue's start time. We dedupe candidates within 90 seconds of each other so you never get a chapter cluster.

Output

  • 5–12 chapters per recording (capped to keep the player chip rail readable).
  • Minimum 90s spacing between chapters.
  • Chip rail rendered below the video — click any chip to seek.
  • Stored in the same recording metadata as the WebVTT URL; no extra processing step.

When chapters don't show up

If there's no transcript, or the transcript has no recognised transition phrases, the chapter chip rail is hidden — we don't show empty UI. Add a transcript (or let the auto-transcription service finish) and reload the player to pick them up.

The chapter parser is heuristic, not semantic. If your lecturing style doesn't include explicit transitions, expect fewer chapters — that's the system being honest rather than inventing fake structure.

Related