Key Takeaways
If the video is yours, YouTube Studio will export the caption track, but it gives you SubViewer (.sbv), not SubRip. The two formats hold the same thing, so converting is mostly a matter of adding cue numbers and changing the punctuation in the timestamps. If the video isn’t yours, the captions are somebody else’s work, and whether you can reuse them is a licensing question about that video, settled before any tool enters the picture.
The request usually arrives sounding trivial. Someone wants the subtitles off a video, as a file, so they can re-cut the piece, hand it to a translator, or put the same captions on the same video somewhere that isn’t YouTube. The text is right there on screen. It should be a two-minute job.
It often is. But the two-minute version only exists in one of the two situations you might be in, and the other one has a question attached that no downloader, extension or converter answers on your behalf. So it’s worth being precise about which situation you’re actually in before you start clicking.
Start by asking whose video it is
Everything downstream depends on this. If the video sits in your own channel, the caption track is yours, YouTube will hand it over, and the rest of the work is a format conversion. If the video belongs to someone else, the captions are their work product, whether a human wrote them or the platform generated them, and your right to reuse them is governed by the licence on that video and not by whether a tool will produce a file.
People conflate the two because the technical route looks similar from the outside. In both cases you end up holding a text file with timestamps in it. But one route is an export from an account you control, and the other is copying material you didn’t make. A tool that produces a file has told you nothing about whether you’re allowed to publish it.
There’s a middle case worth naming: you’re working on a client’s video, or a video your team uploaded under a different account. That’s still “your own video” in the sense that matters, as long as someone can log into the channel. Get the export done from Studio rather than scraping the public page, because the export is cleaner and you’ll be certain you have the track that’s actually attached to the video.
Exporting the caption track from your own video
YouTube Studio will export a caption track directly. You go to the video, open its subtitles, and download the track that’s attached. There’s no reconstruction involved and nothing is being guessed at; you’re getting the same cues the player displays, with the same timings.
What surprises most people is the file that comes back. YouTube Studio exports SubViewer, with an .sbv extension. That’s not the format your editor, your player or your translator’s tooling is expecting. If you’ve only ever handled subtitles as .srt, the download looks broken, and the instinct is to assume the export failed or that you picked the wrong menu item. You didn’t. SubViewer is simply what Studio gives you.
SubViewer is a real subtitle format and it holds everything you need. It’s just older and less widely supported than SubRip, so it tends to bounce off software that’s perfectly happy with SRT. If you want the longer version of the download route, including what to do when the track you want isn’t the one selected, there’s a guide to downloading YouTube subtitles as SRT that covers it.
One habit worth forming: keep the original .sbv after you convert. It costs nothing, it’s a plain text file, and if the conversion goes sideways or you need to redo it with different settings, you’d rather not go back to Studio for a second export.
From SubViewer to SubRip
The gap between the two formats is smaller than it looks. Both are plain text. Both store a list of cues, each cue being a start time, an end time and the words that appear between them. Neither carries styling you’d miss.
SubViewer writes a cue as a timing line with the two timestamps separated by a comma, then the text underneath. It doesn’t number its cues, and it writes hours without padding them out to two digits. SubRip wants a slightly different arrangement: a cue number on its own line, then a timing line written hh:mm:ss,mmm with a comma before the milliseconds, then the text, then a blank line before the next cue.
Which means the conversion is genuinely mechanical. You’re adding sequential numbers, adjusting how the timestamps are punctuated and padded, and inserting the blank lines. The timings and the text carry over exactly. Nothing is interpreted, nothing is dropped, and there’s no quality loss to worry about, because there’s no re-encoding happening — it’s the same words at the same moments, written down differently.
You can do it by hand for a very short video, and people have written regex one-liners for it, but for anything of normal length it’s not a good use of your afternoon. A browser-based SBV to SRT converter does the rewrite and hands the file straight back. Whatever you use, open the result afterwards and look at the first cue and the last cue. First cue confirms the numbering and the timestamp shape are right; last cue confirms nothing got truncated partway through.
When the video belongs to someone else
Settle the rights question first, because it’s the one that can’t be fixed later. The captions on a video you didn’t upload are part of that video’s content. Reusing them — republishing the subtitles, putting them on your own cut, distributing a translation — is a licensing matter that depends on how the video is licensed and what you intend to do. Quoting a line in an article is not the same as shipping a subtitle file with someone else’s video. Personal reference, study notes and research sit differently again from anything you publish. That’s a judgement about the specific video, and it’s made before you touch a tool.
Once that’s actually settled, extracting the text is straightforward. The SozAI Chrome extension exports the transcript of the video you’re watching as TXT, SRT or Markdown, with the timings preserved in the SRT version — so you can skip the SubViewer step entirely and get the format you wanted in the first place. It’s one option among several, and it doesn’t decide the licensing question for you either.
The practical difference between the two routes is worth knowing. The Studio export is the authoritative track, exactly as it’s attached to the video. Anything captured from the watch page is the transcript the player is serving you, which is the same content but reached from the outside. For your own videos, prefer the export. For everything else, the export isn’t available to you anyway.
Fixing what’s wrong without starting over
A subtitle file is plain text. This is the most useful thing about the format and the most consistently forgotten. If one cue has a typo, or a timing is off by half a second, or a name is spelled three different ways, you open the file in any text editor, fix the line, and save. You don’t regenerate anything. You don’t re-run a conversion. You edit a line of text like you’d edit a line of text.
That changes how you approach cleanup. Rather than hunting for a tool that produces a perfect file first time, assume you’ll do one editing pass on the file itself, and get comfortable doing it. Most of what needs fixing is small and local: a proper noun, a stray line break, one cue that lingers.
Structural problems are harder to spot by eye, because the file is long and the errors that break players are usually boring ones — a cue whose end time comes before its start, two cues that overlap, a number out of sequence, a malformed timestamp that survived a conversion. Running the file through an SRT validator before you ship it will surface those faster than scrolling. Then fix them in the text editor, since the file is still just text.
What automatic captions will not give you
If the track you exported was generated automatically, you don’t have a finished subtitle file. You have raw material. Automatic captions carry no punctuation you can rely on, so sentences run into each other and the reader has to do the parsing that a comma or a full stop would have done for them. They also carry no speaker changes, which is a real problem the moment there’s more than one person talking — an interview turns into an undifferentiated stream, and a viewer following the subtitles can’t tell who said what.
None of that stops the file from converting cleanly. SubViewer to SubRip works exactly the same whether the captions were written by a person or generated by the platform, because the conversion only touches numbering and punctuation in the timestamps. It won’t repair the text, and it isn’t supposed to.
Which means a file built from automatic captions needs an editing pass before it goes on anything published. Read it through, add the punctuation, mark the speaker changes in whatever convention you’re using, correct the names and jargon that were always going to come out wrong. On a short video that’s a few minutes. On a long interview it’s real work, and you should plan for it rather than discovering it at the end.
The other thing to expect: line breaks in an auto-generated track are placed for the player’s timing, not for reading. Cues split mid-phrase in ways that look fine scrolling past on YouTube and look careless burned into a video elsewhere. Rebalancing those is part of the same pass, and it’s the part most people skip and then notice later.
