How to transcribe TikTok videos: the three ways, and which one to use

TikTok rarely ships a caption track you can save, which is why "just give me the transcript of this TikTok" turns into a small project. There are three ways that work. Which one fits depends on how many videos you have and how much you want to fiddle.

Why TikTok is harder than YouTube

YouTube captions almost every upload automatically and lets tools read the track. TikTok does not. Auto-captions exist on some videos, but creators switch them on per upload and most never do. Text a creator adds in the editor or in CapCut is drawn into the video, so there is no file behind it. And TikTok blocks automated access aggressively, which is what breaks the do-it-yourself scripts the moment they leave a laptop.

So a transcript of a TikTok usually means the audio, transcribed.

Option 1: by hand

Play the video, turn on captions if the creator added them, and type along. A screen recording helps for anything longer than a minute. Fine for one video. Hopeless past three, and it depends on captions the creator may never have enabled.

Option 2: do it yourself with free software

If you are comfortable with a terminal, two free tools do the job: one downloads the audio from the link, the other turns speech into text.

yt-dlp -x --audio-format mp3 "https://www.tiktok.com/@user/video/7137723462233555205" -o audio.mp3
whisper audio.mp3 --model small --output_format srt

That works on your own computer. Three things make it stop working when you rely on it.

• Blocking. TikTok challenges and refuses requests from servers quickly. What works from home stops working anywhere automated.
• Speed. Speech recognition takes real processing time per clip. A few videos are fine. A few hundred is an afternoon.
• Everything around it. Checking for captions first, timestamps, subtitle formats, retries when a download fails. All yours to build.

Option 3: paste the link

The box on the front page does all of the above in one step. Paste a public TikTok link. It checks for a caption track first, because when one exists it is instant and exact. When there is none, which is most of the time, it transcribes the audio. Either way you get the same clean text, with timestamps where the video has them, and subtitle downloads. No account, no software, and it works on the videos that never had captions.

Photo-mode posts and slideshows work too, as long as someone is speaking. Live replays work once TikTok has saved them.

Which one to use

• One video, once. Any of the three. The paste box is the fastest.
• A handful of videos now and then. The paste box.
• You enjoy the terminal and have time. The do-it-yourself route is genuinely capable. Budget for blocking and processing.

What to expect from the output

Accuracy on TikTok audio is usually good, because most creators talk close to the phone. Three cases come back with nothing or with rough text, and none of them is a bug.

• Music only. A dance clip over a song has no words. That is the honest answer, not something to retry.
• Words typed onto the video. Not speech, so not transcribed. Your phone's text recognition on a screenshot reads those.
• Several people talking at once. Duets and group videos produce text, but check it.

Get started

Copy the link from the share menu, paste it into the box on the front page, and read the TikTok instead of watching it.