How to Download YouTube Auto-Generated Captions

Published on September 19, 2026

You can download YouTube auto-generated captions the same way as captions a creator typed by hand: paste the video link into the transcript.you Subtitle Downloader and save the result as TXT, SRT, VTT, PDF or DOCX. The download is the easy part. Auto-captions come out as short, loosely punctuated fragments with the odd misheard word, so this guide also shows you how to spot them and how to clean them up.

What auto-generated captions are

When a video is uploaded, YouTube runs speech recognition on the audio and, for supported languages, creates a caption track automatically. Nobody reviews it. It sits next to any captions the creator uploaded or typed, which are usually called manual or "standard" captions.

Most videos with clear speech in a supported language end up with an auto-caption track, even if the creator never touched captions. That is why you can usually get a transcript of a random lecture, vlog or interview. The trade-off is quality: the words are mostly right, but the formatting is not written for reading.

How to tell if a video only has auto-captions

YouTube labels auto-generated tracks in the player:

1. Open the video on YouTube.
2. Click the gear icon (Settings) in the player, then Subtitles/CC.
3. Look at the list. An auto-generated track shows as something like English (auto-generated). A track without that label was added by the creator or a contributor.

If the only entry has "(auto-generated)" after it, the video only has auto-captions. If you see both "English" and "English (auto-generated)", the creator also supplied a proper track, and that one is almost always cleaner.

This check matters because our language dropdown shows the tracks a video has but does not say which ones are auto-generated. Check the YouTube player first, and you'll know what you're getting before you download.

How to download auto-generated captions

1. Copy the video's URL from the address bar or the Share button. Regular videos, Shorts and live stream replays all work. A bare video ID also works.
2. Open the Subtitle Downloader and paste the link.
3. The transcript page opens with the full captions, clickable timestamps that jump the embedded player to that moment, a search box, a timestamps show/hide toggle, a Copy button and an Export menu.
4. If the video has two or more caption tracks, use the language dropdown to switch between them. Manual and auto-generated tracks both appear there. If the language you asked for isn't available, the page falls back to the first track the video has.
5. Open Export and pick a format.

Which format to pick

If you're not logged in, you can still read the transcript and use Copy, but you need a free account to download files. Guests and free accounts get one YouTube transcript every 24 hours. Paid plans are unlimited and faster. If you're not sure which file type you need, see SRT vs VTT vs TXT.

What auto-captions actually look like

The SRT and VTT files keep YouTube's own caption timing exactly as it is, with one cue per caption line. With a manual track, a cue is usually a full sentence or a tidy half-sentence. With auto-captions, YouTube splits speech into short rolling chunks, so you get many small cues. A typical stretch looks like this:

1
00:00:00,480 --> 00:00:02,560
so today we're going to look at

2
00:00:02,560 --> 00:00:04,720
how the the new model handles

3
00:00:04,720 --> 00:00:07,040
um long documents and whether it

4
00:00:07,040 --> 00:00:09,200
actually beats the the older one

The same passage in a TXT download drops the numbers and timings but keeps the fragments, one per line:

so today we're going to look at
how the the new model handles
um long documents and whether it
actually beats the the older one

The usual quirks

• Little or inconsistent punctuation. Sentences run into each other, and capital letters are often missing.
• Fragments, not sentences. Cue breaks fall wherever the recognizer paused, often mid-phrase.
• Misheard words. Names, brands, jargon and accents cause the most errors. "Kubernetes" may come out as "cooper netties".
• Filler and repeats. "Um", "uh" and doubled words ("the the") are often transcribed as spoken.
• Sound tags. You may see markers such as [Music] or [Applause].
• No speaker labels. Auto-captions don't say who is talking.

None of this is a download problem. The file matches what YouTube shows on screen.

How to clean up auto-generated captions

How much cleanup you need depends on what the file is for.

For reading, notes or quotes (TXT or DOCX)

1. Pick the right file. TXT has no timestamps. DOCX and PDF put a time marker at the start of each line, which is handy if you want to cite times.
2. Download DOCX if you plan to edit, because it opens straight in Word or Google Docs.
3. Use find and replace to fix repeated errors. If a speaker's name is misheard the same way twenty times, one replace fixes all of them.
4. Search for fillers ("um ", "uh ") and delete them.
5. Add paragraph breaks where the topic changes. The clickable timestamps on the transcript page help here: jump to a spot in the video, listen and check the wording.
6. Fix punctuation last, once the words are right.

Before you quote someone, check the exact wording against the video. Auto-captions are fine for notes but not reliable enough to quote without checking.

For subtitles (SRT or VTT)

Here the goal is readable on-screen lines, not just correct words.

• Fix the words first in any plain text editor. Edit only the text lines, never the numbers or the --> timing lines.
• Merge short fragments into sentence-length cues. Take the start time of the first cue and the end time of the last one. Subtitle Edit and Aegisub, both free, make this much easier than doing it by hand.
• Keep lines readable, roughly 42 characters per line and no more than two lines per cue.
• Save as UTF-8 so accented characters survive.

Our step-by-step guide to editing an SRT file covers merging cues, shifting timings and fixing encoding in detail.

If it's your own video

If you own the video file, you have another option. Upload it to video-to-SRT, which transcribes the audio with AI (Whisper) and gives you a new subtitle file. Compare it with YouTube's auto-captions and keep whichever needs less fixing. If your video has no captions at all, see what to do when a YouTube video has no subtitles.

Common mistakes

• Downloading the auto track when a manual one exists. Check the YouTube player's Subtitles/CC menu, then choose the matching language in our dropdown. The manual track is usually the better one.
• Assuming the transcript is a quote. Check names, numbers and technical terms against the audio.
• Breaking the SRT structure while editing. A missing blank line between cues or a changed arrow can stop players from reading the file.
• Expecting translated auto-captions. YouTube's "auto-translate" subtitle option is created live in the player, and we don't download it. Paid plans can AI-translate the transcript text into 130+ languages, but the result is plain text, not a timed SRT. More in downloading subtitles in another language.
• Reposting someone else's captions. Using them for study, notes or accessibility is generally fine. Republishing another creator's transcript needs their permission.

For the full walkthrough, including YouTube's built-in options and phone versus desktop, see the main guide on how to download YouTube subtitles.

FAQ

Can I download auto-generated captions from any YouTube video?

You can if the video has an auto-caption track. Most videos with clear speech in a supported language do. Very new uploads, music-only videos and videos where the owner turned captions off may have none. In that case the Subtitle Downloader shows an error instead of a transcript.

Why doesn't the language menu say "auto-generated"?

The dropdown lists the caption tracks the video has, but it doesn't mark which are manual and which are automatic. Check the Subtitles/CC menu in the YouTube player, where auto tracks are labelled.

Is the timing in the SRT accurate?

The timing is YouTube's own, unchanged. It lines up with the video the same way the on-screen captions do. The cues are just short, because auto-captions are split into small fragments.

Can I get auto-captions with proper punctuation?

Not from YouTube's auto track itself. You can clean the text up by hand as described above. If it's your own video, you can also create a new subtitle file from the video with video-to-SRT.

Do I need an account?

You don't need one to view and copy a transcript. You need a free account to download TXT, PDF or DOCX, and your first SRT or VTT download is free with that account.