Generate YouTube chapters from video transcripts

How the generator works

When you submit a link, two things happen in sequence, both handled by Marqly's server:

• Fetching the transcript. The server reads the video's caption track from YouTube and converts it into timestamped segments — lines of speech, each anchored to the second it starts at. It prefers the uploader's own caption track and falls back to auto-generated captions when that is all there is.
• Finding the sections. A language model reads the timestamped transcript and decides where the natural breaks are, producing four to ten chapters with short titles. The model only picks boundaries and names: every timestamp in the output is taken from a real caption segment, and any candidate time that does not match the transcript is discarded. The list is ordered by time and forced to start at 00:00, because that is YouTube's hard requirement.

Because nothing is invented, the result cannot point at a minute that does not exist — if a chapter title is slightly off, the time it points to is still a real moment in the video.

How to add chapters to a YouTube video

1. Paste the video URL above and press Generate chapters.
2. Wait for the two passes — reading the transcript, then finding the sections.
3. Review the list — edit any title that misses the point of its section.
4. Press Copy chapters.
5. Paste the timestamps into the video description, each on its own line. On your own video that means the description field in YouTube Studio; on someone else's, keep the list with the save in your library.

YouTube's chapter rules and timestamp formats

YouTube turns a plain list of timestamps into clickable chapters automatically — but only when the list satisfies every rule below. A single violation and the whole list renders as inert text, which is the most common reason "my chapters don't work" threads exist.

Use cases

• Creators publishing long talks, tutorials, or podcast episodes — a chapter list that took an hour of scrubbing to build by hand now takes one paste.
• Students and researchers noting down lectures — the chapter titles map an hour-long lecture into its arguments, and the timestamps let you jump back to the exact passage when you cite it.
• Sharing clips inside a team or family — a chapter list at the top of a description tells people where "the part about pricing" starts instead of making them scrub for it.
• Triaging a watch-later pile — chapters reveal whether a 90-minute video has ten minutes you actually need.

Why chapters are worth the two minutes

• Viewers stay longer — people who can see the structure jump to the part they need instead of leaving.
• Search sees the sections — chapter titles describe the video in words, which helps it surface for more specific queries.
• They double as notes — a chapter list of someone else's talk is the fastest summary of its argument.

Privacy: what happens to the link you paste

The work is server-side, and it is worth being precise about what that means. The video URL you enter is sent to Marqly's own server (running on Cloudflare's infrastructure), which requests the caption track from YouTube and runs the chapter detection there; the transcript segments are processed to produce your result and returned to your browser. There is no account to attach anything to, and no video list, transcript, or result is saved against you. Because the requests are anonymous but expensive, the endpoints are rate-limited per IP — if you are told to wait a minute, that is the limit working, not a failure. The one thing to know: this is a public-web tool. It can only read a caption track that YouTube serves publicly; if the video is private or its captions are not published, there is nothing to fetch and the tool says so.

Tips, limits, and edge cases

• Videos without captions fail. No caption track, no transcript, nothing to chapter — the error says so. Uploaders who disabled captions are out of reach.
• Very short videos are refused. If the transcript is only a handful of segments, the tool reports the video as too short to chapter usefully rather than inventing structure out of noise.
• Long videos are partially covered. Extremely long transcripts are analyzed only up to a point; the output then says so, and the chapters cover the first portion. Re-run on the rest, or edit by hand.
• Auto-generated captions lower the ceiling. Section titles are only as good as the words they are derived from — an ASR track full of mishearings produces vaguer titles than a human transcript does.
• Language handling is pragmatic. The tool looks for an English caption track first and otherwise takes what the video offers, so a non-English video returns chapters titled in its own language.
• Edit freely — but keep the rules. Start at 00:00, keep three or more, ten seconds apart at minimum, and your edited list will still render.

Chapters, summary, or transcript?

Chapters give you the shape of a video; asummary gives you its argument; thefull transcript gives you the exact words to quote. All three come from the same caption track, and all three are free here. If you want a single deep-link into one section rather than a full table of contents, thetimestamp link maker builds a shareable URL that starts at a chosen moment. If you find yourself running videos through these tools regularly, the Marqly extension does it on the watch page itself — and saving a video keeps its transcript attached, sosemantic search can find it months later from a half-remembered detail.

Related tools and guides

• Pull the full transcript of the same video when you need exact quotes, not just section titles.
• Summarize the video instead of skimming its chapters.
• Build a link that starts at one moment — for sharing a single section rather than the whole table of contents.
• Save videos with the Marqly extension so the transcript travels with the bookmark.
• A watch-later system that tells you which half of a 90-minute talk you actually need.
• Marqly's free plan — get started and keep your next video save searchable by what it said.