TikTok captions: the three things people mean, and how to get the words out

Ask five people for a TikTok's "captions" and you will get three different things back. That mismatch is why so many caption tools leave you with the wrong text, or none. This guide separates the three, explains why most TikToks have no subtitle file at all, and shows how to get the words out regardless.

The three things people call "TikTok captions"

The post caption. The text the creator typed under the video, hashtags and all. It is a description, not what was said, and it is the only one TikTok hands over freely. Copying it is a long-press away and needs no tool.

Text on the video. Words the creator placed on the frames, in TikTok's editor or in an app like CapCut. These are pixels. There is no file behind them, and the only way to read them by machine is optical character recognition on each frame, which no transcript tool does.

A real caption track. A timed subtitle file, the same idea as a .srt, made by TikTok's auto-captions or uploaded by the creator. This is the one worth having and the one that is usually missing.

Almost every "TikTok subtitles" tool that disappoints people is quietly giving them one of these when they wanted another.

Why most TikToks have no caption track

TikTok added auto-captions in 2021, and creators switch them on per video. Two things follow from that.

Auto-captions are opt-in for each upload, not automatic. YouTube generates captions for nearly everything, which is why YouTube transcript tools feel effortless. TikTok does not, so on a random TikTok the odds of a track are low.

Text a creator added in CapCut does not create a track either. The video is readable to a person watching it and completely opaque to anything that looks for a subtitle file. The words are on screen and unreachable at the same time.

So if a tool only looks for a caption track, it will come back empty on most TikToks. Nothing is broken. There is simply nothing to download.

When a caption lookup fails, the reason matters

Three failures look alike from the outside and mean very different things.

No captions. The video is public, it has audio, and no caption track was ever made. This is the ordinary case on TikTok. Transcribing the audio will work.

Captions disabled. A track may exist but is not exposed. Transcribing the audio still works.

No speech. There is nothing to transcribe. A dance clip over a licensed song, a silent product montage, a montage with music only. No tool will produce words, because none were spoken.

The first two are worth a second try by a different method. The third is a dead end, and a good tool tells you so rather than spinning.

Getting the words when there is no track

The order that works:

1. Use the caption track if there is one. It is instant, it is exactly what TikTok or the creator produced, and it is already timed.
2. Otherwise transcribe the audio. The speech is pulled from the video and turned into text by a speech recognition model. On TikTok this is the main path, not the backup, because so few videos carry a track.

That is what the paste box on the front page does. Paste a public TikTok link and it checks for a track, transcribes the audio when there is none, and gives you the same clean text either way, with timings where the video has them, so you can export subtitles as well as plain text. No account, and it works on videos that never had captions.

What to expect

Treat the caption track as a bonus and the audio as the real source. Expect music-only videos to come back with nothing, because there is nothing there. And if the words you need are written on the frames rather than spoken, no transcript tool can help: that is a screen-reading problem wearing a captions costume.