One Short in Korean and English: caption styles and rules
Posting a Korean and an English cut of your own AI Short? YouTube is silent on translations; Sume caption styles are not. Which style fits which script.

If you make a Short in two languages from your own script, the October 2, 2026 update (9to5Google, read 2026-10-05) is aimed at content re-uploaded from other creators, and a version of your own work in another language is not that. None of the YouTube pages we read mention translated versions. So the questions that remain are practical, and the main one is the caption style.
What the caption docs require
Sume's caption docs, read on October 5, 2026, say that if you do not send style, the text of the captions sets it: slam for Latin text and black-outline for Korean text. If you pick slam, punch or tiktok-green and send Korean text, the API returns a 400 with caption_hangul_text_latin_style instead of rendering boxes where the glyphs should be. Those styles use Latin display faces without Hangul.
For Korean speech the docs list styles such as black-outline, weight-shift, highlight, pill-karaoke, clip-wipe and editorial-emphasis, plus korean-ad for the ad karaoke look.
| Script language | Safe choice | Will return 400 |
|---|---|---|
| English (Latin) | slam, punch, tiktok-green, or omit style | Hangul faces such as pretendard on a Latin style |
| Korean | black-outline (default), clip-wipe, korean-ad | slam, punch, tiktok-green |
Make the two cuts feel written for each audience
A direct translation tends to carry the first language's rhythm. Rewrite the hook for the second audience, and change an example that does not travel. Two cuts with different hooks are better for viewers and look less like a copy.
Each caption job is $0.20 for videos up to 60 seconds, so two language cuts of a 45-second Short are $0.40 in captions.
Keep the language hint honest
The language field only tells speech-to-text which language to expect. It never selects the style or the font. If your script is authored, send it as script_text so the captions keep your wording rather than a transcript's guess.
Treat the two versions as two Shorts that share an idea. Label any realistic synthetic scene in each, since disclosure is per video.
A worked example
Say you have a 40-second Short about a kitchen tool. The English cut opens with a question and uses slam. The Korean cut opens with a number, uses korean-ad, and passes the Korean text as script_text so the burned words match what you wrote. Both are captioned in one job each, $0.20 apiece, $0.40 in total.
If you ever try the English style on the Korean script by accident, the 400 error costs you nothing, because the docs say a bad look fails at request time and you do not pay for an incorrect render.
Sources
Related posts
More in Use cases
- One-take walkthrough prompts: store, kitchen, studio on Seedance 2.5
Three one-take walkthrough prompts for Seedance 2.5 (store aisle, restaurant kitchen, studio), each with a route, a pace and a 30 s price at 480p to 1080p.
- Open Enrollment Reminder Video With an AI Avatar: Script, Cost
Script an HR open-enrollment reminder as a Sume Avatar 1.0 clip: scenes, word and second counts, request JSON, and the cost on each quality tier.
- Open house invitation clip: MiniMax H3 Max, 8 seconds at 768p
A listing photo to an 8-second 16:9 open-house invitation with minimax-h3-max on Sume costs $0.80 at 768p, $0.50 at 480p and $1.60 at 1080p.
- Open house teaser: four Omni room clips, 360p drafts then 720p finals
A 40-second open-house teaser from four listing photos on Sume: image-to-video clips drafted at 360p, finished at 720p, joined on a timeline. Cost per pass.
Written by Sume