XT-14 // Features
Auto chapters
The video's structure, even when YouTube has none.
Search hits on the progress bar
Many videos ship without chapters. Xtrakt deduces them locally from the transcript: when the vocabulary shifts, a new chapter begins. Each one is titled with its two most distinctive words — no server, no AI service, any language.
Chapters show up as a compact list in the widget — one click jumps straight there — and as thin ticks on the progress bar. When the audience replay curve is available, boundaries snap to the nearest dip. The Markdown export gains one heading per chapter, ready for your AI assistant.
- Detected locally from the transcript — any language
- Clickable list in the widget, discreet ticks on the bar
- Boundaries refined by the audience curve when available
- One ## heading per chapter in the Markdown export