YouTube Chapters From a Transcript: Timestamps Fast

Follow the format YouTube requires

Chapters only appear when the list in your description meets a few rules. Get any of them wrong and the platform silently ignores the whole list, which is the most common reason people think chapters are broken.

• The first timestamp must be 0:00, even if the intro is only a few seconds long
• You need at least three timestamps, listed in ascending order
• Each chapter must be at least 10 seconds long
• Put each timestamp at the start of its own line, followed by a space and the title

Use minutes and seconds such as 4:35, or hours for long uploads such as 1:02:10. YouTube may also generate automatic chapters; your own list in the description takes priority, and you can turn automatic ones off in the upload settings if they pick awkward break points.

Get a timed transcript to work from

You need text with times attached, not a plain block of words. There are three practical ways to get it, depending on whether you have the original file:

• Your own recording on disk: drop it into AI Studio, which transcribes on your device, and export the TXT with timestamps
• An upload with captions: the Show transcript panel under the description lists every line with its time
• Your own upload without captions: download it from YouTube Studio's Content page, optionally extract the audio with Audio Converter, then transcribe that file in AI Studio
• Your own channel: YouTube Studio lets you download your caption track as an SRT file with timings

The timestamped TXT export is the easiest format for this job, because each line starts with a clock time such as [00:04:35] followed by the words spoken at that moment. Copying from the Show transcript panel is quick too, but make sure timestamps are switched on in its menu before you copy; without them you get the words alone, which is fine for reading but not for chapter work.

Spot the topic shifts in the text

Read the transcript quickly, looking for the sentences where the speaker moves on. These are usually easy to see in text even when they are subtle on screen. Watch for signposting phrases such as next, let’s move on, the second thing, a quick recap or now that we have. Each of those is a candidate chapter boundary.

Aim for one chapter every three to eight minutes in a typical tutorial or interview. A 40-minute upload with six to ten sections is easy to scan; 25 sections turn the list into noise. Merge tiny segments into their neighbor rather than creating chapters of 15 or 20 seconds.

If you want a first pass, paste the transcript into the AI Content Pack and use the Chapters button. It proposes an outline of short section titles in order, which you then match to times by searching the transcript. The Content Pack uses a cloud AI model and reads roughly the first 4,500 characters, so for a long upload paste it in parts.

Write chapter titles people actually click

Titles are small but they do a lot of work. Viewers scan them in the list under the player, and Google can show them as key moments. Good ones describe the outcome of the section in three to six words.

• Prefer specific over generic: Export settings for 1080p beats Settings
• Use the words viewers search for, such as reset password, not your internal jargon
• Keep them parallel in style, for example all starting with a verb
• Skip clickbait; a title that promises more than the section delivers costs trust

Before publishing, click each timestamp in the description preview to check it lands on the start of the right sentence. Shift any that cut in mid-word by a second or two so the chapter opens cleanly.

Step-by-step

Common mistakes to avoid

Pro tips

Frequently asked questions