Make YouTube Chapters from Subtitles
Or paste subtitle text
Options
Milliseconds of silence between one cue ending and the next one beginning. Reach it and a new chapter opens; stay under it and the two cues are read as the same chapter. The default of 10000 is not a round number by accident — it is the shortest a chapter may be on YouTube, and because a chapter runs until the next one opens, keeping the box here means every chapter except the last lasts at least ten seconds. Lower it and you get more chapters and shorter ones; at 3000 you can produce marks YouTube will ignore. At 0 every cue becomes a chapter of its own. The pause is measured against the last cue that had words in it, so a blank cue between two spoken ones neither opens a chapter nor breaks one.
Which frame rate the timings are counted in. On the format pages this is what MicroDVD needs, because it stores timing as frame numbers. On the frame rate pages it is the rate the file was timed against — the version the subtitles were made for — and the page's second box takes the rate of the video you are going to use them with. 23.976, 24, 25, 29.97, 30, 50, 59.94 and 60 are the common ones. Getting it wrong shifts the whole timeline, and on the frame rate pages it shifts it by a ratio rather than by a fixed amount.
This page reads a subtitle file and hands back a chapter list: one line per chapter, a start time and a title, in the shape a video description expects. Drop the file in, run it, and the list comes out ready to paste. Nothing is uploaded, so a client's cut or an unreleased episode stays on your own machine.
YouTube's chapter feature has three rules, and this page meets two and a half of them. The first timestamp has to read 0:00, or the whole list is ignored — and since a file's first cue often starts a few seconds in, behind a title card or a moment of silence, the first line is written as 0:00 whatever the file says. Each chapter has to run at least ten seconds, which the default pause of 10000 ms takes care of by construction: a chapter runs from its own start to the next chapter's start, and the next chapter begins at least ten seconds after the last thing said in this one. The remaining half is the count. Three chapters minimum is a property of the file, and this page will not manufacture one by cutting a paragraph in two — a file with two pauses in it gives three chapters, and a news bulletin with sixty pauses gives sixty.
Chapters are laid down by silence rather than by cue. A subtitle file cuts one spoken sentence into three or four cues so each fits its own timing, so a chapter per cue would give you a list hundreds of lines long. Instead the page measures the gap between one cue ending and the next one starting, and opens a chapter when that gap reaches the number in the pause box. Anything shorter is read as the same chapter still going, which is why the words of a later cue can be missing from the list while being perfectly intact in your file: a chapter line is one line, and its title is whatever the first cue of that chapter says.
Times are written the way people write them by hand — minutes and seconds under the hour, and the hour spelled out once a chapter passes it, so a single list can hold 20:00 and 1:02:03 side by side. Both are read correctly, and writing every line the long way would turn a 20-minute mark into 0:20:00, which is harder to scan and says nothing more. Seconds are cut down rather than rounded up, so a time never points at a moment that has not arrived yet.
Cues with no text in them are skipped and leave no line of their own, but the time they occupy still counts as silence: the gap is measured from the last cue that actually had words, not from a blank one that happens to sit between them, because what a viewer hears is the silence. Two things this page will not do are join your subtitles into readable prose, which is a separate page, and put speakers' names on lines, because none of the formats read here has a field for who is talking.
Examples
A file that gives three chapters
Input
1
00:00:02,000 --> 00:00:05,000
The room is dark
and the heating is off.
2
00:00:05,400 --> 00:00:07,000
Nobody has been here
since Tuesday.
3
00:00:22,000 --> 00:00:25,000
A cold night,
and no heating.
4
00:01:01,000 --> 00:01:04,000
Every night this week?
Output
0:00 The room is dark and the heating is off.
0:22 A cold night, and no heating.
1:01 Every night this week?
Four cues in, three chapters out. The first two cues are 400 ms apart, so the second one stays inside the first chapter and does not appear in the list at all — the words are still in the file, they just are not a chapter heading. The two later cues each follow a silence far past the ten-second threshold and open chapters of their own. The first line reads 0:00 even though the first cue starts two seconds in, because that is the only value YouTube accepts there.
A pause of exactly ten seconds
Input
1
00:00:03,000 --> 00:00:06,000
We can talk about it
in the morning.
2
00:00:16,000 --> 00:00:18,000
Every night this week?
Output
0:00 We can talk about it in the morning.
0:16 Every night this week?
The silence here is exactly 10000 ms, the number in the box, and that is the moment a new chapter starts. This is the example to look at if you are checking the arithmetic: raise the box past 10000 and the two cues fold into one chapter, because the gap no longer reaches the threshold. Lower it and the same thing happens to files with less breathing room.
A chapter past the hour
Input
1
00:00:05,000 --> 00:00:08,000
The room is dark.
2
00:20:00,000 --> 00:20:03,000
Nobody came.
3
01:02:03,000 --> 01:02:06,000
A cold night,
and no heating.
Output
0:00 The room is dark.
20:00 Nobody came.
1:02:03 A cold night, and no heating.
Two writing styles in one list, on purpose. A chapter under an hour is written as minutes and seconds (20:00) and one past the hour carries the hour (1:02:03), which is how a person writes a chapter list by hand and both are read correctly. Writing every line the long way would turn the second chapter into 0:20:00 — longer, and no easier to find.
Frequently asked questions
How many chapters will my file give me?
As many as it has long pauses in, which is not a number this page decides. A chapter line is opened wherever the silence between two cues reaches the pause box, so a half-hour interview with steady speech may well give you three or four chapters, while a news bulletin with a pause between every item gives you dozens. If the result is too coarse, lower the pause and run the file again; if it is a wall of lines, raise it. There is no setting that turns a file with two pauses into a file with ten.
Why does the list start at 0:00 when my subtitles start later?
Because a chapter list whose first timestamp is not 0:00 is discarded entirely rather than read from the second line on, and a file's first cue very often starts a few seconds in — behind a title card, a logo, or a moment of room tone. The first chapter covers the opening of the video either way, so 0:00 is a true description of it, and it is the only value that makes the list work. Every later line keeps its own real time, so the shift applies to the first line and to nothing else.
What happens if I lower the pause below ten seconds?
You get more chapters and you lose the guarantee that each one lasts ten seconds, which is YouTube's own minimum. The ten seconds come from the pause setting rather than from a separate rule: with the box at 10000, a chapter cannot be shorter than the gap that opened the next one. At 3000 the same file may produce chapters of four seconds, and YouTube will ignore chapter marks that short. The box goes down to zero, where every cue becomes its own chapter — useful for checking what is in a file, less so for publishing.
Where did the rest of the words go?
Nowhere — they are still in your file, and this page never writes to it. A chapter heading is one line of text, and the page uses the first cue of each chapter for it, so anything said after that cue is inside the chapter rather than in its title. If what you want is every word as readable prose, that is a different page: the transcript export joins the cues into paragraphs and keeps all of them.
Why do some lines read 20:00 and others 1:02:03?
Because that is how chapter lists are written by hand, and a player reads both without trouble. A mark under an hour is minutes and seconds; once a mark passes the hour the hour is written out so the number still makes sense at a glance. Writing every line the long way would give 0:20:00 for the first of those, which is longer, no more precise, and slower to scan when you are looking for a section.
Do I already need three chapters for this page to be useful?
The page is useful either way, but it cannot create a third chapter out of a file that only has two pauses in it. Inventing one would mean cutting a sentence in half and calling the second half a new section, which would put a chapter mark in the middle of a thought and give you a list that no longer matches the video. If your file genuinely has only two breaks, the honest answer is that it has two chapters' worth of structure, and the fix is a shorter pause or a different file.
Can I paste the list straight into the video description?
Yes. That is what it is for: plain text, one line per chapter, a time then a title, no numbering and no punctuation added anywhere. Paste it on its own lines in the description and the chapter marks appear. The one thing to keep is the first line exactly as it comes out — if you delete the 0:00 or move that chapter later, the list stops being recognised, however sound the rest of it is.