How to Download a YouTube Transcript (and Make It Usable)

Every captioned video on YouTube has already been transcribed. The machine did the work, the text exists, and it's sitting three clicks away from you. The frustrating part isn't getting it — it's that what comes out is almost, but not quite, the thing you wanted.

You went looking because you needed the words. A two-hour conference talk with one useful segment somewhere in the middle. A competitor's product walkthrough you want to quote accurately. Forty minutes of lecture that has to become revision notes by Thursday. Whatever the reason, scrubbing the timeline with your mouse is not a research method, and you know it.

So here's how to download a YouTube transcript on desktop and on your phone, what to do when the button that's supposed to be there isn't, and — the part most guides skip entirely — how to turn the raw output into something you'd actually reread.

Three routes out, in the order most people should try them

Which one you want depends on a single question: do you own the video?

If you're a viewer: the transcript panel

On desktop, open the video, expand the description, and click Show transcript. A scrolling panel opens to the right of the player with the full text, broken into timestamped lines that highlight as the video plays. Select it, copy it, paste it wherever you're working.

That's the whole method, and it's the part worth being blunt about: there is no download button for viewers. YouTube shows you the text and expects you to take it with Ctrl+C. Every "YouTube transcript download" tool you'll find is doing some version of this on your behalf — fetching the caption track and handing it back as a file.

If you own the video: YouTube Studio

Creators get the thing viewers don't — an actual file. In YouTube Studio, go to Subtitles, pick the video, and use the Download menu to export .srt or .vtt. These are proper caption formats with timing data intact, which matters if you're feeding them into an editor or re-uploading captions elsewhere. If the transcript is destined for anything other than a text document, start here.

If neither of those works: a transcript tool

Paste the video URL, get the text back. These exist because of the two gaps above — no viewer download, and nothing at all on mobile. The good ones also run speech recognition when a video has no captions to fetch, which is the one thing YouTube itself cannot do for you.

Quick tip: Once the transcript panel is open, use Ctrl+F (Cmd+F on Mac) to search the text, then click the matching line. The player jumps straight to that moment. It turns a two-hour video into something you can navigate in about fifteen seconds, and you never have to copy anything.

When the button that should be there isn't

This is the most common thing that goes wrong, and the explanation is simpler than the forum threads suggest. YouTube's transcript is generated from captions. No captions, no transcript — there's nothing underneath the feature to display.

So when Show transcript is missing, one of these is true:

• The creator turned captions off. Their choice, and there's no way around it from the viewer side.
• The video is too new. Automatic captioning runs after upload and takes a while. Come back later.
• The language isn't covered. Auto-captioning supports a long list, but not everything.
• A browser extension is interfering. Ad blockers and YouTube enhancers reshuffle the page and sometimes bury the control.

There's a fast way to tell the first three apart from the fourth: look for the CC icon in the player. If it's absent, no caption track exists and no tool on earth can fetch one — the only remaining option is running speech recognition on the audio itself. If CC is there but the transcript option isn't, the page is at fault, not the video.

Quick tip: Open the same video in a private or incognito window before you go hunting for a fix. Extensions don't load there, and site data can't interfere. If the transcript appears, you've found your culprit in about five seconds.

The app will read it to you but won't let you take it

Here's the one that sends people to Google at 11pm. On the YouTube mobile app, you can tap the video title or …more to expand the description, and on many videos a transcript is there. You can read it. You can follow along. What you cannot do is select the text and copy it.

It's a strange limitation, and it has no workaround inside the app itself. Two things do work:

1. Request the desktop site. In Safari or Chrome on your phone, open the video, then use the browser menu to switch to the desktop version. The transcript panel appears as it would on a laptop, and text selection comes back with it.
2. Copy the link and paste it into a transcript tool. Tap Share, Copy Link, then paste it into whatever tool you use. This is the route most people land on, and it's why "youtube transcript" tools get so much mobile traffic — they're solving an app limitation, not a technical one.

If you're on a phone and in a hurry, the second option is faster. The desktop-site trick is more fiddly than it sounds on a small screen.

What 90% accuracy actually costs you

Anyone who's read auto-captions closely has a favourite mangled sentence. Creators have a name for them: craptions. But the problem deserves real numbers, because "pretty good" and "good enough to quote" are very different standards.

YouTube's automatic captions land between 85% and 95% accurate, and the variable is almost entirely the audio, not the algorithm. A clean studio recording with one person speaking clearly hits 94–96%. The same system on video shot outdoors with traffic and wind drops to 78–82%.

Now do the arithmetic. People speak at roughly 150 words a minute, so a 40-minute video is about 6,000 words. At 90% accuracy, that's 600 wrong words scattered through your transcript. Not gibberish — mostly small things: a missing comma that reverses a clause, a name spelled four different ways, a technical term replaced by whatever rhymed. Enough to make a direct quote risky and a published excerpt embarrassing.

The practical rule: auto-captions are fine when you're searching, skimming, or summarising. They're not fine when a specific sentence is going into an article, a citation, or a legal file. For those, read the passage you're quoting against the audio before you publish it. It takes a minute and saves a correction.

Nobody's name is in the file

Search for how to get speaker names into a YouTube transcript and you'll find a lot of workarounds and no real answer. The real answer is that YouTube's automatic captions have no concept of speakers at all. The output is one continuous stream of text. A three-person panel discussion comes back looking exactly like a monologue.

For a solo explainer, who cares. For an interview, a podcast episode, or a recorded meeting, it's the difference between a usable document and a wall you have to reconstruct from memory. You end up scrubbing the video anyway — which was the thing you were trying to avoid.

What fixes this is speaker diarization: the transcription separates voices and tags each one. Most paid transcription tools include it, and tools built for meetings — Laxis among them — treat it as the default rather than a feature, because a meeting transcript without speaker labels isn't much of a transcript. The labels still arrive generic (Speaker 1, Speaker 2), so you rename them once at the top and the whole document becomes readable.

Quick tip: Whatever tool you use, spot-check the transcript at the points where speakers change. Diarization is most likely to slip during crosstalk, and an error there doesn't look like an error — it looks like someone said something they didn't.

The twenty minutes between raw text and something you'd reread

A transcript is raw material. Copying it into a document and calling the job done is where most people stop, and it's why so many transcripts get downloaded once and never opened again.

Four things make the difference, and none of them take long:

Drop the timestamps if you don't need them. Inside the transcript panel, the three-dot menu has a Toggle timestamps option. Turn it off before you copy and you get continuous prose instead of a numbered list you'd otherwise strip by hand.

Fix the proper nouns first. Names, products, and jargon are where speech recognition fails hardest and where errors do the most damage. A find-and-replace on the three or four terms that matter cleans up most of the mess in under a minute.

Add the paragraph breaks. Auto-captions arrive as one undifferentiated block. Breaking it at topic changes is what turns a transcript from something you search into something you read.

Write the summary while it's fresh. Three bullets at the top — what this was about, what was decided or claimed, what you want to do with it. Future you will read those bullets and skip the rest, which is the entire point.

That last step is where most of the value lives, and it's the one people skip because it's the only part that takes thought. It's also what the better transcription tools now do for you. If your recording is a meeting rather than a YouTube video, that workflow is largely automated — Laxis captures the call, labels who spoke, and produces the decisions-and-action-items version alongside the full text, so the summary exists before you go looking for it. Different source, same principle: the transcript is the input, not the output.

The transcript was never the goal

Notice what's happened to this task over the past few years. Getting words out of a video used to be genuinely hard — you paid someone, or you typed it yourself at four hours per hour of footage. Now the text is pre-generated, free, and three clicks away, and the entire difficulty has migrated downstream to shape: no speaker names, no paragraphs, no summary, and a copy button that mysteriously doesn't work on the device most people are holding.

Which suggests the thing to optimise isn't extraction. Almost nobody actually wants a transcript. They want to find one moment in two hours, or check a quote, or turn a talk into notes — and the transcript is just the least-bad format that stands between them and that. The tools worth paying for in 2026 aren't the ones that fetch text fastest. They're the ones that understand the text was never the point.

Frequently asked questions

How do I download a YouTube transcript?

On desktop, expand the video description and click Show transcript. A timestamped panel opens beside the player; select the text and copy it. YouTube gives viewers no download button, so copying is the download. If you own the video, YouTube Studio hands you a real file: Subtitles, pick the video, then Download as .srt or .vtt.

Why is Show transcript missing on YouTube?

The transcript is built from captions, so no captions means no transcript. That happens when the creator disabled them, when the video is too new for auto-captioning to finish, or when the language is not yet supported. If the CC icon is absent from the player, there is no caption track to read. Browser extensions can also hide the button — opening the video in a private window rules that out in a few seconds.

How do I get a YouTube transcript on my phone?

The YouTube mobile app can show a transcript under the expanded description, but it will not let you select or copy the text. To take it, request the desktop site in your mobile browser and use the transcript panel there, or copy the video link and paste it into a transcript tool. Copying inside the app is the one thing it does not allow.

How accurate are YouTube auto-captions?

Between roughly 85% and 95%, and the variable is the audio rather than the algorithm. Clean studio recordings reach 94–96%; video shot outdoors with background noise drops to 78–82%. At 90% accuracy a 40-minute video carries around 600 wrong words, which is why auto-captions are a starting point rather than a finished transcript.

How do I remove timestamps from a YouTube transcript?

Open the three-dot menu inside the transcript panel and choose Toggle timestamps. The time markers disappear and the text becomes continuous prose you can paste into a document without stripping numbers line by line.

Do YouTube transcripts include speaker names?

No. YouTube's automatic captions are one undifferentiated stream of text with no speaker labels, so an interview or a panel arrives looking exactly like a monologue. Tools with speaker diarization separate the voices and label them Speaker 1 and Speaker 2, leaving you to attach the real names once.