How to add subtitles to TikTok: in-app text, auto captions or burn-in

Decide what a subtitle means inside TikTok

TikTok's editor is built for speed, and its text tools reflect that. You can add a text layer, style it, and time it on the timeline; some app versions also offer automatic captions. A text layer is not the same as a subtitle track. It is a design object placed over the video, usually burned into the exported file, and it does not give the viewer a separate caption control. That distinction matters when a client asks for subtitles and expects something a viewer can switch off.

For most TikTok posts, the practical choice is between an in-app text layer and a burn-in from an external editor. A sidecar SRT is useful upstream: it carries the timing and wording into the editor, and it can be translated or reviewed. It is not a delivery format that TikTok documents for an ordinary post. Build the SRT as the master, then choose how the words reach the viewer.

TikTok subtitle requirements: what is actually documented

TikTok's help centre describes adding text and captions in the video editor, but the pages we checked do not publish a file-format specification for caption uploads on standard posts. That is different from YouTube, which lists SRT, WebVTT and other caption formats. The absence of a public spec is not proof that no import path exists anywhere; it means you should not promise a client that an SRT will attach, and you should check the current help page and your own app before building a workflow around it.

The safer reading is that the in-app tools are the documented route and that availability can vary by region, app version and account. Automatic captions, in particular, are a feature that may appear for one creator and not another. Build the text so it works as a burn-in if the in-app route is unavailable. That way the post still ships, and the SRT remains useful for the next platform.

In-app captions versus an SRT versus a burn-in

An in-app text layer is fast and native. It inherits the app's fonts and animations, sits in the editor where you can nudge it, and exports with the video. The cost is control: precise timing, line breaks and consistent styling across a series are harder than in a desktop editor, and a long caption may need to be split by hand. Auto-generated text also needs a review pass for names, numbers and slang.

An SRT plus a burn-in gives you repeatable typography and timing. You can review the cues in a timeline, apply one style to the whole track, and export a clean master. The trade-off is an extra app and an extra encode, but for a series the SRT is the better investment because the tenth video costs less than the first.

Generate short-cue SRT text before you open the editor

Generate the transcript with video to SRT, then open it in the SRT editor and shorten every cue. TikTok viewers are often reading while the edit is already moving, so a cue that contains a full sentence with two clauses will either be rushed or linger over the next shot. Split at the clause, keep the key word early, and remove repeated filler that the ear does not need.

Do not translate the platform's pace into a rule that every cue must be under one second. A short exclamation can be readable in 800 ms, while a product name with a number may need two seconds even after trimming. The test is whether you can read the line once, comfortably, while watching the clip. If you have to read it twice, it is too long or too slow.

1. Generate the transcript and listen to the opening three seconds.
2. Split sentences at clause boundaries, not at scene cuts.
3. Shorten each cue to one or two short lines.
4. Check names, products, numbers and negations.
5. Export UTF-8 SRT and keep the master for the next cut.

Safe area and framing for a vertical TikTok

The TikTok interface covers the bottom of the frame with the username, caption and sound line, and the right edge with the action buttons. Those overlays move as the app changes, so treat any pixel measurement as temporary. The durable rule is to keep the words in the central band, leave a margin from the right edge, and keep the lowest line above the area where the app's own text sits.

Vertical framing also changes what centred means. A caption centred in the video file can look low once the interface is drawn over it, and a caption placed high can collide with a title or a creator's face. Preview the composition with the app's overlays visible, not only in the editor's clean viewer. The subtitle line length habits that work for other vertical platforms are a useful starting point, but this app's layout is the final check.

• Keep the caption block in the central band.
• Leave the bottom area clear for the username, caption and sound line.
• Leave the right edge clear for the action buttons.
• Test the longest product name, not the shortest word.
• Preview with the interface visible, not only in a clean viewer.

Cue timing for fast speech and hard cuts

Fast speech is not the same as fast reading. A creator may deliver a sentence in two seconds, but the viewer still needs time to parse the words, especially if the sentence contains a name or a number. Start the cue on the stressed syllable, end it after the thought, and avoid changing the text in the middle of a gesture that the viewer is following. A hard cut is a natural place for a cue change; a cut in the middle of a word is not.

When two speakers overlap, do not stack two full captions in the same position. Either combine the exchange into one cue with clear labels, or split the screen position deliberately and keep it consistent. If the overlap is brief, a single cue that captures the meaningful line is often better than two cues fighting for the same space. Check the result at normal speed, not frame by frame.

Editing auto-generated text without losing the beat

Automatic captions are a draft. Read them once with the sound on and once with the sound off. The first pass catches words the recogniser misheard; the second catches lines that are technically correct but split in a way that makes the sentence hard to follow. Pay particular attention to product names, handles, numbers, and negations such as not and never, because those change the meaning of the post.

If the app lets you edit the generated text, make the correction in the smallest unit possible rather than retyping the whole block. If it does not, export the video and caption it in an editor where you can control the cues. Keep a note of the corrections in the SRT so the next post with the same vocabulary starts from a better transcript.

Export and upload: keeping the text readable

Export the highest-quality master you can, then create the captioned version from that master. A caption burned into an already compressed file inherits the compression, and small text suffers first. Keep the caption style simple: one weight, one colour, a readable outline or background, and a size that survives a phone screen at arm's length. Test the busiest frame, not a clean title card.

If you are uploading a file with burned-in text, watch the uploaded result rather than the local preview. The app may re-encode the video, and that can soften edges or shift the apparent size of the text. If the result is hard to read, fix the source and upload again rather than adding a second text layer on top. The burn-in guide covers the export settings that keep small type intact.

Common failures and a final phone check

Start with the track or layer you can verify. If the words are missing, check whether the text layer was actually included in the exported file, whether the caption tool was available for that account, and whether the viewer has captions turned off. A burned-in caption that disappears is usually an export or layering problem, not a platform bug. Our guide to subtitles not showing covers the file-level checks.

Finally, watch the post on a phone with the sound off, then with the sound on. Check the first second, the fastest exchange, and the loop point. The VoiceToTextFree home page has the generator and editor if you need to rebuild the text, and the captioning workflow is useful when the TikTok post is one part of a larger release.

• The text is visible in the exported file, not only in the preview.
• The caption tool or auto-caption feature was available for the account.
• The viewer's caption setting is not hiding a track.
• The lowest line clears the username and caption block.
• The loop point does not cut off the final cue.