Video Accessibility: Captions, Transcripts & More

Most web accessibility work happens in markup: a missing alt attribute, a low-contrast button, a form field with no label. Those are machine-detectable, which is why automated scanners catch them. Video is different in kind. The barrier isn’t in the code — it’s in the content. A deaf visitor needs the words that are spoken. A blind visitor needs the actions that happen on screen described. Neither of those is something a parser can generate or verify.

This is exactly where the “30–40% of WCAG” ceiling comes from. Automated scanning — ours included, and CompliaScan is built on the open-source axe-core (opens in new tab) engine — reliably covers roughly a third of the success criteria. Nearly all of the time-based media criteria fall in the other two-thirds, the part that requires a human. Anyone who tells you a scanner or a widget can make your videos accessible is selling you something. We wrote a whole piece on where automated testing ends and manual testing begins, and video sits squarely on the manual side of that line.

These words get used interchangeably, and the confusion causes real failures. Getting the distinction right is half the battle:

• Captions assume the viewer cannot hear the audio. They include speaker identification and non-speech sound — [door slams], [upbeat music], [laughter]. This is what accessibility law requires.
• Subtitles assume the viewer canhear but doesn’t speak the language. They translate dialogue and skip the sound cues. A subtitle track is not a caption track, and it does not satisfy WCAG.
• Transcripts are a static text version of the whole thing. They’re the accessibility workhorse: usable by screen readers, searchable, translatable, and indexable by Google.

One more distinction worth the ink: open captions are burned into the video and always visible; closed captions live in a separate file (a WebVTT or SRT track) the viewer can toggle. Closed is almost always the better choice — it lets users control font size and position, and it keeps the text out of your video pixels so it doesn’t pixelate on a small screen.

You don’t need a media accessibility vendor to get this right. A repeatable, mostly-free process handles the vast majority of web video:

1. Auto-generate, then edit. Let your platform produce a first-draft caption file, then have a human clean it up in the platform’s caption editor — fix names, punctuation, and speaker labels, and add bracketed sound cues where they carry meaning.
2. Publish a transcript on the page. Paste the corrected text directly onto the page below the player, or link to it. This single step covers audio-only content, helps many audio-description cases, and hands Google a block of indexable, keyword-rich text it can rank.
3. Script visuals into the narration when you can. The cheapest audio description is the kind you never have to add later: if your presenter says “click the blue Scan button in the top right” instead of “click here,” the visual information is already in the audio.
4. Caption live events in real time. Zoom, Teams, and the major webinar platforms all offer live captioning — turn it on. For higher-stakes events, book a live human captioner (CART).
5. Use an accessible player. Make sure the play, pause, and volume controls work with a keyboard alone and don’t autoplay with sound. This part is machine-testable — a scan will flag an unlabeled control or a keyboard trap in the player chrome.

Captions and transcripts are one of the rare accessibility investments that pay for themselves in plain business terms. A large share of social video is watched on mute; captions are why anyone finishes it. Transcripts give search engines something to index that a video file never could, which is why so much video content quietly underperforms in search — there’s nothing on the page to rank. And translated captions open your content to audiences you currently can’t reach at all.

Where a scan genuinely helps is everything aroundthe video — the player controls, the page structure, the contrast of your caption styling — and in catching regressions when someone swaps a video and forgets the caption file. Run the page through our free WCAG checker, work the machine-detectable issues off the list with the WCAG checklist, and pair it with periodic monitoring so a broken caption track can’t sit live for months. Just keep the mental model honest: the scanner protects the frame around the video, and a human still owns the words inside it. If you want a structured way to check the human-judgment items too, our accessibility audit guide walks through the manual review.