TikTok Spark Ads translation skips text burned into the video

TikTok's captions and translations for Spark Ads auto-generate captions from the video's audio and let viewers translate the ad title, description, subtitles, and interactive-element text, but "Text that is part of the video or image creative itself is not translated", and the page says AI-generated creatives are not supported. For an ad with words burned into the frames, a viewer's translate toggle will not help, so you need a version of the video per market. Sume's video captions burns one language per job.

The TikTok facts come from About Captions and Translations for Spark Ads, read on 2026-10-02.

What does the Spark Ads translation cover?

The feature is viewer-initiated. People see a "See translation" link below the ad description and can toggle between the translated and original version. Neither creators nor advertisers edit the translations; the system is automatic.

It is available in the United States and European Union regions only, according to the page.

What does that mean for burned-in captions?

A caption you burn into the file is part of the creative, so it stays in the source language for every viewer. If your Spark Ad carries burned English captions and a viewer in Germany taps translate, the title and description change but the burned words do not.

That is a reason to keep the source file clean when you plan to localize: put the speech in the audio, let TikTok's own subtitles handle captions where the feature applies, and burn text only where you control the market.

How do you localize burned text per market with Sume?

Take the clean clip, then run one caption job per language. The video captions page lets you pass cues with text, start, and end in seconds, which burns that copy without speech-to-text, or script_text, which keeps the speech-to-text timings and aligns your wording to them. language is only a speech-to-text hint; it does not choose the style or font.

Each accepted job reserves $0.20 for video up to 60 seconds, so four markets cost $0.80 in captions. Each result is its own video, so each market becomes its own ad, and you give each a different Idempotency-Key. The walkthrough in caption one video in three languages shows the pattern.

What about AI-generated creatives?

TikTok says the translation feature does not support AI-generated creatives. If your Spark Ad is built from a Sume avatar video or other generated footage, plan on not getting the viewer-side translation at all, and ship the localized versions yourself.

Whether a given post counts as AI-generated is decided by TikTok's labeling, not by Sume. Read the label rules in TikTok AIGC label or your own caption before you rely on either route.