Skip to content

How Do You Get a TikTok Video Transcript?

You can, when the video carries a caption track. TikTok generates captions with speech recognition, but no public TikTok API returns them to a commercial developer, so the transcript has to come off the public video.

8 min read · 16 Oct 2026

Yes, if the video has a caption track. TikTok generates captions for most spoken-word videos with speech recognition, and that text is attached to the public video. What does not exist is an official TikTok API that hands a transcript to an ordinary developer — the only API field that returns caption text is gated to vetted academic and non-profit researchers. So a working transcript comes from reading the caption track off the public video, not from a developer endpoint.

Here is exactly what TikTok produces, what its APIs will and will not give you, and where the text actually lives.

Does TikTok transcribe its own videos?

It does, and has since 2021. TikTok introduced auto captions on April 6, 2021, starting with American English and Japanese. The feature uses speech recognition to "automatically generate subtitles," and creators "have the power to edit the text of their captions once they're generated." It was built for accessibility — viewers who are "hard of hearing or deaf" — not as a data product, which is why the transcript exists but is not handed out through an API.

A year later, on July 21, 2022, TikTok extended it two ways: "in addition to creators, viewers now have the option to turn on closed captions for videos," and it added machine translation of captions, initially across "English, Portuguese, German, Indonesian, Italian, Korean, Mandarin, Spanish and Turkish," on "select videos."

Two things follow from this that matter if you want the text:

  • A video only has a transcript if TikTok generated one for it. Spoken-word videos usually do; a music-only clip usually does not. There is no transcript to fetch where none was made.
  • TikTok ships both an original-language track and translated tracks. If you blindly take "the first caption track," you can end up with a machine translation of an English video into Spanish. The original spoken track is the one that reflects what was actually said.

Will any official TikTok API return the transcript?

For a commercial developer, no — and this is a common pattern, the same shape as LinkedIn's API returning only the authenticated member's own profile rather than arbitrary lookups. TikTok runs three public API products, and only one exposes caption text at all:

  • The Research API has a field named voice_to_text, documented as "Voice to text and subtitles (for videos that have voice to text features on, show the texts already generated)." It is the only official caption field. But access is restricted: eligible applicants are academic institutions or not-for-profit research organizations that are "independent of commercial interests," and TikTok is explicit that "your developer account alone is not sufficient to grant you access to Research Tools." A product company does not qualify.
  • The Display API returns a creator's video metadata — id, title, description, duration, cover, counts — and documents no caption, subtitle, or transcript field.
  • The Content Posting API is publish-only; it uploads videos and returns nothing about captions.

So the official surface gives transcripts to approved researchers and to no one else. Everyone else is left with the public video.

Where the transcript actually lives

When TikTok has made captions for a video, they travel with the public video page as a caption track — the same track the "Show captions" toggle switches on for a viewer. Third-party tooling has long observed that the public video's data carries a list of caption tracks with their languages and files; those files are standard WebVTT served from short-lived URLs. TikTok does not document this as a supported format, so treat it as "how the public page works," not as an API contract: the shape can change, and the file URLs expire quickly.

If you only need one video, the TikTok transcript extractor does exactly this read and shows you the text, with timestamps or as an SRT file, or tells you the video has no track. It sits with the rest of the free tools, which need no account. Measuring the videos as well as reading them is covered in How Do You Check a TikTok Engagement Rate Without Logging In?.

Reading it reliably means a few things that are easy to get wrong:

  • Pick the original spoken language, not the first track in the list. TikTok orders tracks for its own reasons, and the first one can be a translation.
  • An absent track is an absent track, not an error to paper over. A video TikTok never captioned has no transcript, and the honest answer is "none," not a guess.
  • The file URLs expire, so the text has to be fetched and stored when the page is read, not linked to.

Getting it through TrueScrape

TrueScrape exposes this as a single endpoint, /v1/tiktok/transcript, built around the two traps above. It returns the spoken transcript as one text block plus timed cues, and it defaults to the original spoken-language track rather than whichever track TikTok happens to list first, so the text reflects what was said. You can pass a language code when you want a specific track. When a video has no caption track, the call returns an error that lists what is available instead of silently substituting another language — and, because an empty result is never billed, a video with no transcript does not cost a credit.

curl "https://api.truescrape.com/v1/tiktok/transcript?url=https://www.tiktok.com/@user/video/1234567890" \
  -H "x-api-key: YOUR_KEY"

That is the whole shape of it: TikTok writes the transcript, keeps it off its public APIs, and leaves it on the public video — so getting a TikTok transcript at scale is a question of reading the right caption track, not of finding an endpoint that was never built.

Sources

Everything described here runs on the same API. You are never charged for a failed request, an empty result, or a cache hit.