TikTok Captions vs a Transcript: What's the Difference?

Published on September 23, 2026

Captions are text shown on top of a TikTok while it plays. A transcript is the spoken words written out as text you can copy, search, edit and export as a file. They look similar on screen, but only one of them leaves you with something you can paste into a document, and that difference decides which tool you reach for.

If you want the spoken words as text, paste the link into the TikTok transcript tool and read the result on screen. If you want words burned onto the video for viewers, that is a captioning job and it happens inside a video editor or inside the TikTok app.

The three kinds of text on a TikTok

People say "captions" for three different things, and mixing them up is why so many how-to posts contradict each other.

1. On-screen text the creator added

Text stickers, title cards, punchlines typed into the editor. Once the video is exported, this text is part of the picture — it is pixels, the same as a tree or a face. Nothing in the file says "the word 'sourdough' appears here." It also frequently has nothing to do with the audio: a cooking clip might show "Day 4" while the voice talks about hydration levels.

2. TikTok's auto-captions

TikTok can generate captions from the speech and show them during playback, and viewers can usually toggle them. They are readable in the app. What they are not is a file you can take away — there is no export button that hands you a .srt or a .txt, and you cannot select the words the way you select text on a web page. You can read them; you cannot reuse them.

3. A transcript

A separate text document of what was said, usually with timestamps. It lives outside the video. You can search it, quote it, paste it into an email, translate it, or convert it into a subtitle file.

Why burned-in text can't be pulled back out

This is the question that sends people in circles. If the words are right there on screen, why can't a tool just grab them?

Because by the time a TikTok is published, that text has been rendered into the video frames. Getting it back means optical character recognition — reading the letters out of the image — and OCR on short-form video is unreliable: the text moves, it sits over busy footage, it uses decorative fonts with outlines and drop shadows, and it appears and disappears mid-word during animations. Even when OCR works, you get the title cards, not the speech.

The sturdier route is to ignore the picture entirely and work from the sound. That is what our tool does: it feeds the audio to an AI speech model (Whisper) and writes down what it hears. The upshot is useful — the video does not need captions of its own for this to work. A clip with no on-screen text at all transcribes perfectly well, as long as someone is talking.

The flip side is equally important: if nobody speaks, there is nothing to transcribe. A dance clip over a trending song comes back empty, because music is not speech.

What a transcript gives you that captions never will

• Text you can actually take. Copy the whole thing, or one line, into your notes, a document, a script.
• Search. Find the 12 seconds where the claim you half-remember was made.
• Timestamps. Every line carries a start time, which is why SRT and VTT exports come out as real subtitles with proper timing rather than one enormous block of text.
• File formats. TXT, SRT, VTT, DOCX and PDF from the same result page, via the Export menu.
• Translation. A Translate menu covering 130+ languages, on paid plans, with plain text output.
• A record. Signed in, each transcript is saved to your Files page so you can come back and export it in another format later. As a guest you get the result on screen only.

When you want which

The accessibility angle

Captions and transcripts serve overlapping but different needs. Captions help someone watching the video — deaf and hard-of-hearing viewers, anyone in a quiet office or a loud train, people following a second language. Transcripts help everyone who is not watching: screen reader users who would rather read at their own pace, people who scan before committing three minutes, anyone who needs to reread a dense explanation, and search engines indexing the page you posted it on.

If you publish short video and you care about reach, do both. Add captions so the clip works on mute, and post the transcript alongside it so the content exists as text as well. Our WebVTT and SRT exports cover the caption half; the TXT or DOCX export covers the other.

Getting the transcript

1. Copy the link. In the TikTok app, tap Share, then Copy link. On desktop, take it from the address bar. Both tiktok.com/@user/video/… and vm.tiktok.com short links work.
2. Paste it into transcript.you/tiktok and start it.
3. Read the result, hit Copy, or open the Export menu for a file.

Reading and copying are free with no signup. Downloads need a free account; the first SRT or VTT is free on a free account, after which they are on paid plans, and translation is paid-only. Guests and free accounts get one TikTok transcript per 24 hours; paid plans are unlimited and run with priority. The full walkthrough, including the phone flow and troubleshooting, is in how to get a TikTok transcript.

Honest limits

Accuracy follows the audio. Clear, close-mic speech comes back near-perfect. Loud backing music, two people talking over each other, heavy accents and street noise all produce errors, so read before you quote — a habit worth keeping if you are transcribing TikToks for research or journalism. Private, deleted and region-blocked videos cannot be fetched at all; you get an error rather than a partial result. And transcribing a public video for notes, study, accessibility or research is fine, but republishing someone's words as your own needs their permission.

FAQ

Can I export TikTok's own auto-captions?

No. They are generated and displayed by the app; there is no file to download. To end up with a caption file, transcribe the audio and export SRT or VTT — the steps are in downloading TikTok subtitles as an SRT.

Does the video need captions for your tool to work?

No, and that is the point. It reads the audio, not the screen, so a clip with no captions and no on-screen text transcribes fine. What it does need is audible speech.

Will it pick up the text stickers on screen?

No. Those are part of the image. If a title card says something the voiceover never says, it will not appear in the transcript.

What about a video that is only music?

It comes back empty. There is no speech to write down. Lyrics sung over a track are hit-and-miss at best and should not be relied on.

Is a subtitle file the same as a transcript?

Almost. A transcript is the words; a subtitle file is the same words cut into short timed chunks a player can display. Because each line here already has a start time, the SRT and VTT exports are produced from the same transcript.