· Everpop
The best caption style for YouTube Shorts is white, sentence-case sans-serif text at a line height of 3.9–4.5% of frame height (BBC), with a thick dark stroke or box, in one or two lines timed to speech, between the speaker's face and the player's buttons. Burn it in; add a caption track for screen readers.
What is the best caption style for YouTube Shorts?
The best style for YouTube Shorts captions, by broadcast and accessibility guidelines, is white sans-serif text set to a line height of 3.9% to 4.5% of frame height (BBC), at least 4.5:1 contrast against its backing (WCAG), one or two lines per card (DCMP), and timing that starts with the speech.
| Property | Guideline (1080×1920 Short) | Source |
|---|---|---|
| Font | Plain sans-serif; distinct I, l and 1; open counters; medium weight | DCMP, Braille Institute |
| Size | Line height 3.9–4.5% of frame height (about 75–86 px, by arithmetic); "the largest size needed"; bigger is a style choice with fewer characters per line | BBC |
| Lines | One or two (the BBC allows three on 9:16), each up to 90% of width, about 25 characters | DCMP, BBC |
| Colour | White; yellow, cyan or green for another speaker | BBC |
| Backing | Shadow, box or a halo-wide stroke; contrast at least 4.5:1 against what sits behind the letters | DCMP, WCAG 1.4.3 |
| Active-word highlight | Optional; in one small study of language learners (n = 49), synced keyword highlights were preferred for learning but seen as too distracting to replace standard captions in everyday viewing; if used, a subtle, steady colour change; on a word box, dark text | arXiv study, Disabled World, WCAG formula |
| Case | Sentence case; caps for a stressed word | DCMP, BBC |
| Timing | At speech onset; ideally, with the speaker in shot, within 1.5 s of speech either side; at least around 0.3 s per word | BBC |
| Position | BBC limit: central 75% of height, 90% of width; on Shorts (Google ad template), x 48–888 px, y 288–1,248 px | BBC, safe-zone guide |
Under WCAG 1.4.3, a wide border that "fills in the inner details of the letters" acts as a halo and "would be considered background", so a thick black stroke keeps white text readable over a white kitchen counter. A hairline is not a halo: "A narrow border around the letter would be used as the letter." By the same page's formula, white on a yellow (#FFFF00) word box computes to about 1.07:1 (black, about 19.6:1), and the BBC no longer uses coloured backgrounds.
How do you test a caption font for YouTube Shorts?
A caption font for YouTube Shorts passes the legibility test when its look-alike letters, such as I, l and 1, stay distinct at phone size. BOIA recommends "simple, unadorned fonts that are widely available such as Arial, Lucida Sans, Tahoma, Verdana, and Helvetica". The case for burning captions in for muted viewers, and for a plain, licensed face, is made in burned-in captions for muted Shorts.
Two principles behind the Braille Institute's Atkinson Hyperlegible, a free typeface for low-vision readers, work as tests: "Similarly-shaped upright letters are distinctly different", and counters, the areas inside letters, "keep characters clear". Type Illinois and the digit 1 at final size on a phone; if the uprights blur, pick another face. DCMP asks for "medium weight"; if you go heavier, keep those counters open.
What size should YouTube Shorts captions be?
The BBC's subtitle guidelines set caption line height for 9:16 video at 3.9% to 4.5% of frame height, roughly 75 to 86 pixels on a 1080×1920 Short by arithmetic, and call it "the largest size needed for presentation". That is line height, not an editor's font-size number. It is the BBC's 16:9 range of 7% to 8% scaled by 9/16, because "8% of a vertical video is much larger than 8% of a landscape video".
Going bigger is your call, not the BBC's, and fits fewer characters per card; check it on a phone at arm's length.
How high should captions sit on a YouTube Short?
On a YouTube Short, captions should sit in the band between the speaker's chin and the player's bottom text, left of its button rail. The BBC notes that 9:16 subtitles often sit "a little higher up, though still generally in the lower third", because "faces are generally in the top half of the screen".
Our Shorts safe-zone guide sets that band's floor and side margins from Google's vertical-ad template, which is conservative for organic Shorts. So on Shorts the caption band ends just above the lower third, not inside it; the Position row carries the pixels.
Word-by-word or sentence captions: which reads better?
Word-by-word (karaoke-style) captions win on sync; sentence captions win on reading comfort. A short phrase on screen with a subtle highlight on the spoken word is the middle path. In one small study of language learners (n = 49), synced keyword highlights were "the preferred design for learning" but were "perceived as too distracting to replace standard captions in everyday viewing scenarios" (arXiv 2307.05870).
The BBC says "standard prepared subtitles should be displayed in blocks", and with per-word reveals "viewers cannot read ahead" (Disabled World).
Which caption style fits podcasts, gaming and tutorials?
On YouTube Shorts, podcasts suit one caption colour per speaker, gaming suits small boxes clear of the HUD, and tutorials suit blocks held longer, away from the demo.
| Content | Caption style | Trade-off |
|---|---|---|
| Talking head | Phrase on screen, subtle highlight on the spoken word (or none), under the chin | A highlight can distract; bigger type means more cards |
| Two-voice podcast | Phrase blocks, one colour per speaker, on a dark backing | Word-by-word over crosstalk is hard to follow |
| Gaming | Small boxes wherever the HUD is not, even high, inside the spec's Position margins | Moving captions costs consistency |
| Tutorial or screen recording | Blocks held longer, away from the demo | On-screen interface text competes |
| Music-heavy | Word-by-word or cumulative lines on the rhythm; ♫, not an emoji (BBC) | Rhythm wins over reading comfort |
The BBC backs two rows: a speaker "should keep that colour" (host white, guest yellow), and when the action sits low, as in snooker, "top screen positioning will be a more acceptable standard".
How do you make Shorts captions accessible?
To make Shorts captions accessible, keep contrast at 4.5:1 or better, cover nothing that matters, and upload a caption track too: burned-in captions cannot be resized, switched off or read by a screen reader (BOIA). WCAG 1.2.2 accepts open captions as a sufficient technique; the track adds viewer control.
Pre-publish checklist:
- 4.5:1 contrast against the footage, or a stroke thick enough to count as background
- No face, mouth, on-screen text or HUD covered
- Sentence case, lines broken at natural points
- Names, brands and jargon proofread before the burn
- Each caption lands with the voice
- Highlight, if any, limited to "a subtle and steady change" (Disabled World)
- A caption track uploaded alongside the burn
Where Everpop fits. Everpop is an AI clipping tool that turns your own uploaded videos into YouTube Shorts, Instagram Reels, Facebook Reels and TikTok posts. It burns word-by-word captions into clips, the sync-first side of the trade-off above, and by default nothing posts until you approve it, so every clip can be checked against that list before it goes out. Every plan has 6 curated fonts, Pro and Scale add your own brand font, and captions come natively in 9 languages, with 90+ others captioned in English. In the editor, a clip's first 3 re-renders are free, and any clip exports to Final Cut Pro, Premiere Pro or DaVinci Resolve as FCPXML, EDL and SRT, which gives you an SRT for the caption track (pricing). Agents on Pro or Scale turn burned captions on or off per request with captionsEnabled on POST /api/v1/clip (OpenAPI spec).
Claims in this article
| Claim | Source |
|---|---|
| 9:16 size and its scaling, width, lines, colours, backgrounds, position, snooker, timing, breaks, blocks, caps, ♫ (px by arithmetic) | BBC |
| Contrast, halo, narrow border, formula (ratios computed); open captions | WCAG 1.4.3, WCAG 1.2.2 |
| Weight, backing, case, lines; uprights, counters; font list, burn limits | DCMP, Braille Institute, BOIA |
| Keyword-highlight study (n = 49); reading ahead, subtle highlight | arXiv 2307.05870, Disabled World |
| Shorts margins (Google ad template); band ends above the lower third | safe-zone guide |
Everpop captions, fonts, languages, SRT, re-renders, approval, agent access on Pro/Scale, captionsEnabled |
pricing, agents, OpenAPI spec |
Frequently asked questions
- Should Shorts captions have a stroke or a box?
- A stroke and a box both work on Shorts captions, as long as the text holds contrast. WCAG treats a wide border that fills in the inner details of the letters as a halo, which counts as background, so a thick dark stroke keeps white text readable over bright footage. A thin outline is not a halo: WCAG counts "a narrow border around the letter" as part of the letter, not as background. DCMP prefers a drop or rim shadow and a translucent box, especially on light backgrounds. Either way, aim for at least 4.5:1 contrast against what sits behind the letters.
- Should Shorts captions be in all caps?
- Mostly no; keep Shorts captions in sentence case. The DCMP Captioning Key prefers mixed case for readability, and the BBC warns that large chunks of capitals are hard to read. Keep caps for a stressed or shouted word, where they carry meaning.
- How big should Shorts captions be in pixels?
- The BBC's guidelines for vertical 9:16 video set caption line height at 3.9% to 4.5% of frame height, roughly 75 to 86 pixels on a 1080×1920 Short by arithmetic, and call that the largest size needed. That is line height, not an editor's font-size number. Bigger captions are a style choice that fits fewer characters per line, so check them on a phone at arm's length.
- What colour should captions be on YouTube Shorts?
- Use white text on a dark stroke, shadow or box for Shorts captions. The BBC's default is white on black, with yellow, cyan or green kept for other speakers. WCAG's minimum contrast for normal text is 4.5:1, so check the backing holds that ratio over the brightest part of the footage.
- Is yellow a good caption colour for YouTube Shorts?
- As text on a dark backing, yes: yellow comes second after white in the BBC's order of speaker colours, and by WCAG's contrast formula pure yellow on black computes to about 19.6:1. As a box behind white text, no: that pairing computes to about 1.07:1, far below the 4.5:1 minimum. If you use a highlight box at all, put dark text on it; the BBC no longer uses coloured backgrounds.
- How many words should a Shorts caption show at once?
- A Shorts caption should show only as much as fits on one or two short lines. The BBC's default is one complete sentence per subtitle, split at clause or phrase boundaries when it runs long; at its recommended size a 90%-wide vertical line holds about 25 characters, and the Shorts safe area is narrower, so on Shorts that usually means a short sentence or a phrase. For example, "Nobody tells you that the hard part of a podcast is the editing" becomes three cards: "Nobody tells you", "that the hard part of a podcast" on two lines, and "is the editing". DCMP prefers no more than two lines per caption. The BBC advises against splitting an article from its noun. Bigger type means more, shorter cards.