Many pages that promise a TikTok transcript ask for a link and then return whatever caption track TikTok has stored for the video, which is nothing at all when the creator did not switch captions on. This page does not fetch anything from TikTok. You bring the video file, and the transcript is made from its sound by a speech recognition model, the same way any other uploaded recording is handled.
That takes one extra step on your side, saving the video, and in return it does not depend on captions existing, on the video being available in your region, or on TikTok changing how its pages work. It also means the result has punctuation and sentence boundaries instead of a run of caption fragments.
In the TikTok app, the Share arrow on the right of the video opens a sheet with a Save video button. The file is written to the gallery or camera roll with a TikTok watermark and the creator's username on it. On a computer, tiktok.com usually offers a download option when you right-click the video. Both exist only when the creator allows downloads; the setting is theirs, and many accounts turn it off.
When there is no save option, use the screen recorder built into the phone: Screen Recording in the iPhone Control Centre, or Screen record in the Android quick settings. Start it, play the video once from beginning to end with the volume up, and stop. The result is an ordinary video file with the clip's sound in it.
For your own videos there is a cleaner source. The original clip is normally still in your camera roll or editing app, and TikTok can save a copy to the device when you post. Please only save and transcribe videos you made or have the right to use; a transcript for your own notes or research is a different thing from reposting someone else's work.
A person talking to the camera transcribes cleanly, including the fast, clipped delivery that is normal on TikTok. Narration made with the app's text-to-speech voices is just as easy for the model, because the voice is clear and steady. A voice over a quiet backing track is usually fine; a voice competing with a loud trending sound loses words.
Two things are outside what a transcript can do. Text that appears on screen, such as stickers, burned-in captions and the list items people point at, is not read: only speech is transcribed. And sung lyrics come out poorly, so a lip-sync or a dance video with a song and no talking will give you very little. The spoken language is detected automatically, and more than fifty languages are supported.
Creators turn their own videos into captions, blog paragraphs and posts for other platforms without typing them out again. Marketers collect the opening lines of videos that performed well and compare how the hook is worded. Researchers and journalists quote exactly what was said and keep the text alongside the saved file, which matters when a video is later deleted.
Others simply want the recipe, the workout order or the list of book titles from a clip that goes by too quickly to write down. With the text on screen you can search it and copy the part you need.
Without an account you can transcribe 10 recordings of up to 3 hours each, with 5 hours in total, which is far more than 10 short videos will use. A short clip is shown in full as the free preview. A free account keeps your transcripts in a history and adds an editor; it shares the same allowance and does not add hours. Subtitle export as SRT, VTT or ASS, translation, AI summaries and export to PDF or Word are part of the paid plans.
The speech models run on Transcrify's own servers, and the audio is not sent to a third-party speech service. A video uploaded as a guest is deleted after a day, while the transcript stays available.