How to Add Captions to YouTube Shorts (3 Ways)

There are three ways to caption a YouTube Short: let YouTube generate automatic captions and correct them in YouTube Studio, upload your own SRT subtitle file, or burn open captions into the video before you post. Automatic captions cover Shorts as well as long-form video, but YouTube tells creators to always review them.

The fastest route is not automatically the best one, and the three are not interchangeable — they behave differently for viewers, for accessibility, and for the muted scroll. Here is what each one actually does, checked against YouTube's Help documentation as of October 2026 rather than memory.

What counts as a Short, and why do captions matter more there?

On standard channels, for uploads on or after 15 October 2024, YouTube categorises any video with a square or vertical aspect ratio running up to three minutes as a Short (YouTube Help). That means your captioning decisions happen in the same YouTube Studio surface you already use — but the viewing context is different, and that is what changes the answer.

Shorts are consumed in a scrolling feed, frequently on a phone, frequently with the sound off. Survey evidence points the same way, with the usual caveat that stated preference is not measured behaviour. A Verizon Media and Publicis Media study of 5,616 US adults aged 18–54, run in April 2019, found 80% said they were more likely to watch an entire video with captions, and 50% agreed captions are important and that they usually watch videos with the sound off on mobile and desktop (3Play Media). A later poll points the same way: in a YouGov poll that Stagetext reported in March 2023, 61% of 18- to 24-year-olds said they generally prefer subtitles on when watching TV shows or films in their native language (Stagetext).

There is also a floor underneath the retention argument. Under WCAG 2.2, providing captions for prerecorded audio in synchronised media is Success Criterion 1.2.2 at conformance Level A — the lowest, most basic tier — and the W3C lists both open (always visible) captions and closed captions as sufficient ways to meet it (W3C). Captions are not a growth tactic that happens to help deaf and hard-of-hearing viewers. They are the baseline, and any retention effect is the bonus.

How do YouTube's automatic captions work on a Short?

YouTube's speech recognition creates automatic captions for both long-form videos and Shorts, in every language on its published list — 67 of them as of October 2026, from Afrikaans to Zulu — and YouTube publishes them on the video automatically once they are available (YouTube Help). They are not instant: YouTube says processing time depends on the complexity of the video's audio, and quality varies with mispronunciations, accents, dialects and background noise. YouTube's own instruction is blunt: "Always review automatic captions and edit any parts that haven't been properly transcribed" (YouTube Help).

To review and correct them:

1. Sign in to YouTube Studio.
2. From the left menu, select Subtitles.
3. Click the Short you want to caption.
4. Under "Subtitles," click More next to the subtitles you want to edit.
5. Review the automatic captions and edit or remove any parts that haven't been properly transcribed.

To change caption text or timing, select Edit in the "Subtitles" column for that language — for an automatic track you will first be asked to Duplicate and edit, then Continue — then click into any line to fix the text, adjust the two timestamp boxes beside the caption track, and click Publish (YouTube Help).

If no automatic captions appear at all, YouTube lists the usual causes: the captions aren't ready yet because the audio is complex, the language isn't supported, the video is too long, the sound quality is poor or the speech isn't recognised, there is a long period of silence at the beginning, or multiple speakers overlap or multiple languages are spoken at once (YouTube Help). A Short with loud music under the voice can run straight into the sound-quality one.

How do you upload your own SRT file to a Short?

You add it from the Subtitles page in YouTube Studio, and if you already have an accurate transcript — from your editing timeline, your long-form video, or your script — uploading beats correcting a machine draft line by line.

1. In YouTube Studio, select Subtitles, then the video you'd like to edit.
2. Click ADD LANGUAGE and select your language.
3. Under "Subtitles," click ADD.
4. Choose Upload file (then "With timing" or "Without timing"), Auto-sync (enter the words or upload a transcript and let YouTube time it), or Type manually (YouTube Help).

Format matters. For SubRip (.srt) and the other basic formats, YouTube is explicit: "Only basic versions of these files are supported. No style info (markup) is recognized. The file must be in plain UTF-8." WebVTT (.vtt) supports positioning, but styling is limited to bold, italic and underline (YouTube Help). If you were hoping to control caption colour or font through an uploaded SRT or VTT file, that is not what these formats do on YouTube.

You can also pull a track back out: Subtitles → Edit → Options → Download subtitles gives you a caption file you can fix in a text editor and re-upload (YouTube Help). Once a track is corrected, that is the cheapest way to keep a clean copy of your captions as a file you own and can reuse elsewhere.

How do you burn open captions into a Short?

You render the captions into the video's pixels in an editor or clipping tool before you upload, so they become part of the picture. The viewer cannot switch them off, cannot resize them, and cannot translate them — and they appear the instant the Short starts, with no processing delay and no toggle to find.

For a muted feed, that is the point: the words are on screen before anyone decides whether to keep scrolling. It is the style Everpop renders. Everpop is an AI clipping tool that turns long videos you upload — it never downloads from YouTube — into review-first YouTube Shorts, Instagram Reels, Facebook Reels and TikTok posts, with speech-timed captions burned in.

The part to get right is placement. YouTube's Shorts editor shows visual guides for exactly this reason: the guides "show you where elements like the like button, comments, and video description might appear for viewers," and dragging an overlay too close to the edge makes animated white lines appear, marking a non-safe area where your content might be partially hidden on some devices (YouTube Help). If you burn captions in an external editor, keep them clear of the lower-right stack and away from the very bottom.

Which route is right for your Short?

For most creators it is two routes, not one: burned-in captions for the muted scroll, plus an uploaded or corrected text track for everything pixels cannot do.

Burn captions in for the scroll, and still attach a text track: viewers can change its font, colour and size in YouTube's caption settings (YouTube Help), and the text can be translated and read by machines. Burned-in pixels carry no text layer, so a screen reader has nothing to read. For the full open-versus-closed comparison, see subtitles vs captions.

What should your captions pass before a Short goes out?

Six checks: accurate text, safe placement, readable pacing, a first line timed to the hook, an attached text track, and honest language labels.

• Transcript checked against the actual audio — names, jargon, numbers, brand spellings (why AI captions get names wrong).
• Captions sit inside the safe area, clear of the like button, comments and description.
• Reading speed is survivable: short lines, no wall of text on a quick beat (the best caption style for Shorts).
• The first line lands with the hook, not after it.
• A text track — an SRT, or auto-captions you have corrected — is attached, even if captions are burned in.
• Caption language matches the audio language, and translated tracks are labelled honestly.

Keeping an SRT alongside the rendered clip is also what keeps a workflow portable. Everpop exports SRT together with FCPXML and EDL into Final Cut Pro, Premiere Pro and DaVinci Resolve, so the caption work travels with the footage instead of being locked to whatever rendered it (plans; what each export carries). Whatever tool you use, ask the same question before you commit: can I get the text back out?