Smart+ Automatic Enhancements dub 50+ languages: captions need you
Smart+ Automatic Enhancements can resize, upscale and dub video in over 50 languages. What it leaves to you: on-screen caption text, and how Sume burns it in.

Smart+ Automatic Enhancements can resize, sharpen and translate and dub videos in over 50 languages inside Ads Manager, but TikTok's pages do not say it rewrites text that is already burned into your pixels. If your clip carries on-screen captions, check each market's version before launch and burn local captions yourself; Sume's caption API does that for a public video URL.
This is a different question from the Spark Ads translation caveat covered elsewhere. Here the tool sits in a Smart+ campaign flow and is toggled per capability.
What does Automatic Enhancements include?
TikTok's Symphony automation post says it improves video resolution, cleans up image clarity and resizes assets to fit the vertical screen. It also lists "translating and dubbing videos in over 50 languages" and refreshing music from TikTok's Commercial Music Library. The help article lists the same group: resizing, quality boosts, translation and dubbing. The capabilities are toggleable in the Smart+ workflow.
TikTok reports one advertiser with a 50% lower CPA after using translation and dubbing for an EU expansion. As with other vendor case figures, it is one reported result and TikTok adds that past performance does not predict future results.
What can burned-in text break when you dub?
Dubbing changes the audio track. It does not, on anything TikTok's pages state, change words drawn into the frame: a price tag, a headline card or hardcoded subtitles. A French voice over English on-screen text is a mismatch viewers notice in the first second.
So decide per clip. If the video has no burned-in text, let the tool dub it. If it does, keep a text-free master and burn captions per market after the audio is final.
| Element | Handled by Automatic Enhancements | Your action |
|---|---|---|
| Spoken audio | Translation and dubbing, per TikTok | Review the dub in the preview |
| Burned-in captions | Not stated on TikTok's pages | Burn market captions from a clean master |
| Music | Can refresh from Commercial Music Library | Confirm the track suits the market |
| Resolution and size | Resize and quality boost | Still supply a 9:16 master |
How do you burn market captions with Sume?
Use video captions on a public HTTPS clip URL. To burn exact translated copy without speech-to-text, pass cues with text, start and end; the docs say authored cues skip speech recognition, which also avoids a failure on a silent clip. The tiktok-green style exists for Latin text, and Korean copy needs a Hangul style: sending Korean text to slam, punch or tiktok-green returns a 400.
If you start from a finished spoken clip instead, language is only a speech-to-text hint and never picks the style.
curl -X POST https://api.sume.com/v1/video-captions \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"video_url": "https://example.com/clean-master.mp4",
"style": "slam",
"cues": [
{ "text": "Livraison offerte", "start": 0.0, "end": 2.5 },
{ "text": "Offre limitee", "start": 2.5, "end": 5.0 }
]
}'Which markets are worth a separate caption pass?
Not every language needs a manual pass. Start with the markets that account for the most spend, and for each one check three things in the dubbed preview: that the dubbed line is the same length as the on-screen caption, that numbers and currency match the market, and that brand names are not translated. A caption that runs two seconds longer than the voice reads as an error.
Keep one clean master per clip with no burned-in text. Dubbing, resizing and caption burning then become independent steps you can redo without re-editing the source. If a market underperforms, you can change only the caption cues and re-render, instead of rebuilding the whole clip.
What does Sume not do here?
The Sume docs I read list no dubbing or translation endpoint, so do the audio in TikTok or another tool. Sume can trim a clip to the length you need and inspect it, including optional transcription, to read what is actually spoken before you decide what to translate.
Sume also does not read your Ads Manager settings, so it cannot tell which enhancements you toggled on. Keep a note per campaign of which capabilities were enabled. When a market's results differ, that note tells you whether the audio, the captions or the music changed between the two versions you are comparing, and which of the three to test next.
Finally, check TikTok's own AI-content disclosure options in Ads Manager before you publish dubbed versions; its pages I read do not say how dubbing is labeled, so do not assume either way.
Sources
Related posts
More in Use cases
- Smart+ automatic placement spans Lemon8 and Pangle: one 9:16 master?
TikTok says Smart+ automatic placement can reach TikTok, Pangle, Lemon8 and the Global App Bundle. Start from one vertical master, then check each placement.
- Smart+ Recommended Creatives: web only, so who makes app ads?
TikTok's Recommended Creatives builds video from a destination URL and is limited to Smart+ web campaigns. What to supply yourself and what Sume can make.
- TikTok split test: how to test AI avatar ad hooks
TikTok's split test changes one variable and lists the opening 2-3 seconds as testable. Render each hook as its own Sume avatar video and run 7+ days.
- TikTok split test variables: which ones need a new AI video?
TikTok lets a split test change one variable. Only Creative and Creative Assets need new media; the rest reuse one clip. Here is what to render, with Sume.
Written by Sume