Does YouTube translate Shorts captions? Auto vs your own cues

YouTube makes automatic captions for Shorts, but they are written in the video's default language only. Translation is a separate layer. A viewer can pick Auto-translate in the caption settings when their language is not listed, and you can add your own subtitles in other languages. If you need the translation to read the way you wrote it, you write the lines yourself and either upload them to YouTube or burn them into the video.

The YouTube side below comes from four Help pages read on 2026-10-03: Use automatic captioning, Add subtitles & captions, YouTube tools to translate your content, and Manage caption settings - Android. Sume's side is Video captions.

What does YouTube translate on its own?

Automatic captioning uses speech recognition and, per the help page, works on long-form videos and Shorts. The same page warns that automatic captions might misrepresent the speech because of accents, mispronunciations, or background noise, and tells you to review and edit them. Captions may also not be ready when you upload, since processing time depends on the audio. The subtitles page adds that automatic captions will be in the video's default language only.

The viewer-side translation is on the caption settings page: if the language you want is not listed under Subtitles/CC, click Auto-translate and select a language. That page is about watching, not uploading, and it does not mention Shorts, so I cannot say how it behaves in the Shorts player.

What does your own translated track give you?

The subtitles page describes uploading a subtitle file with timing or without it, adding a language with ADD LANGUAGE, and an auto-sync option that times a transcript to the video. It gives no Shorts-specific steps, so check Studio on a test upload before you rely on it. The translation tools page also lists translated titles and descriptions, which YouTube localized thumbnails and titles walks through.

The advantage of your own file is control: brand names, product terms, and jokes stay as you wrote them. The cost is that you supply the translation, because Sume's documentation lists no translation feature on the caption endpoint.

Where do translated cues made with Sume fit?

Sume's caption job burns text into the picture. Video captions takes a public HTTPS video URL and, if you pass cues with text, start, and end in seconds, burns exactly that copy at those times with no speech-to-text. Each accepted job costs $0.20 for a video up to 60 seconds, and the docs state that SRT uploads are unsupported, so phrase-level cues replace the file. For one translated version per market, Translate an SRT and burn it in shows the converter and Cost of subtitling 20 Shorts in 5 languages prices a batch.

Two cautions from the docs. The language field is only a speech-to-text hint, not a translator and not a style picker. And Latin styles such as slam have no Hangul glyphs, so Korean text on one returns a 400; use a Hangul style such as black-outline with Korean copy. Burned-in text is visible to everyone and cannot be turned off, so it suits a single-market clip; for a Short aimed at many markets, you pay for one burned version per market.

{
  "video_url": "https://media.sume.com/artifacts/artf_demo/clean.mp4",
  "style": "black-outline",
  "cues": [
    {"text": "10초면 충분합니다", "start": 0.0, "end": 2.2},
    {"text": "서랍 하나로 시작하세요", "start": 2.2, "end": 4.8}
  ]
}

Which should you pick?

Let automatic captions run for the original language and review them. For other languages, start with your own translated subtitle file on YouTube, where viewers can switch it on or off. Reach for burned-in cues when the words are part of the design, such as a hook line in the first seconds, and treat dubbing as a later step; Localize a webinar Short argues the same order.