Video caption requirements for business sites

The five media criteria a business site has to meet

WCAG puts audio and video under one guideline, 1.2 Time-based Media, with nine success criteria. Five of them sit at Level A or AA, the levels a business site aims for; the other four are Level AAA. The table above sorts them by the kind of media you publish. Here is what each of the five says, in the words of WCAG 2.2 itself.

Three terms carry most of the weight. Synchronized media is audio and video together, which is what nearly every business video is. Live means sent out as it happens, with no more than a broadcast delay; anything else, including a webinar recording you post the next day, is prerecorded. And an alternative for time-based media is a document that gives the sound and the picture in order, as text: a transcript that also describes what is shown.

WCAG has one standard and three versions; our guide to what WCAG is explains the levels. All five criteria above are the same in WCAG 2.0, 2.1 and 2.2.

What a caption has to carry besides the words

WCAG defines captions as a synchronized text alternative for speech and non-speech audio. Unlike dialogue-only subtitles, they also carry sound effects, music, laughter, and speaker identification and location. W3C's Understanding note for 1.2.2 puts it as three jobs: captions include the dialogue, identify who is speaking, and include meaningful sound effects and other non-speech information.

The Department of Justice says the same in its web accessibility guidance: videos can be made accessible with synchronized captions that are accurate and identify any speakers. Here is the difference on three cues from a physiotherapy clinic's welcome video.

W3C lists leaving out dialogue or important sound effects as a named failure of 1.2.2 (F8). Not every sound needs a tag: traffic outside the window does not, the phone that interrupts the speaker does. Timing counts as well, since captions are synchronized by definition, and WCAG adds that captions should not cover up relevant information in the video, such as a name on screen.

Closed captions or open captions

WCAG accepts both. Closed captions are a separate track the viewer can switch on and off; open captions are part of the picture and always show. For a video on your own site, closed captions in a WebVTT file are the better default: you can fix a typo without exporting the video again, and the player can let the viewer make the text bigger. Open, or burned-in, captions earn their place where you cannot attach a file, which is most social feeds.

The side-by-side table of the two, with the WCAG techniques behind each, is on our captioning software page. Whichever you choose, the content rules above are the same.

A word on “subtitles”: in some countries the word means captions. W3C uses captions for the same language as the speech and subtitles for a translation, and notes that translated subtitles are not in themselves an accessibility feature. A Spanish subtitle track does not caption an English video.

Automatic captions are a draft, not the finished file

“Automatically-generated captions do not meet user needs or accessibility requirements, unless they are confirmed to be fully accurate. Usually they need significant editing.”

W3C Web Accessibility Initiative, Captions/Subtitles

The same W3C page gives an example worth showing to anyone who thinks the machine is close enough. The cook says “Broil on high for 4 to 5 minutes. You should not preheat the oven.” The automatic caption reads “Broil on high for 45 minutes. You should know to preheat the oven.” Two small slips, and the caption tells a deaf viewer the opposite of the video. W3C's advice is to use automatic captions as a starting point and edit them.

YouTube says much the same in its help page on automatic captions: it encourages creators to add professional captions first, and to review automatic ones and edit what was transcribed wrongly. In practice the edit pass is where names, prices, speaker labels and sound tags go in, because a speech engine rarely gets any of them right.

Transcripts: required for audio, not enough for video

A podcast or a recorded message has no picture to caption, so 1.2.1 asks for a transcript instead: the words and the sounds that matter, in order, with the speakers named. W3C's transcripts guide calls a transcript that also describes the picture a descriptive transcript, and notes it is what lets a person who is both deaf and blind follow a video, read on a braille display.

For a video with sound, a transcript is a good extra and a poor substitute. It does not meet 1.2.2, which asks for captions in sync with the video. A full descriptive transcript does meet 1.2.3 at Level A, but at Level AA the video needs audio description under 1.2.5, so a transcript alone no longer meets it. When a video needs description and when it does not is the subject of our guide to what audio description is.

If you already have a caption file, you are most of the way there. Most caption editors export a plain-text transcript; add the speaker names, any text shown on screen, and a line for each visual step, and post it as HTML right under the player.

Where the requirement comes from for a small business

The ADA itself never mentions captions on a website. The Department of Justice has no regulation setting detailed web standards for businesses open to the public; its position is that the ADA's general rules still apply to websites, and it names WCAG and the Section 508 Standards as “helpful guidance”. Its web guidance lists six examples of website barriers, and “no captions on videos” is one of them. The guidance is informal and not legally binding, but it tells you where DOJ looks first.

The same page lists the cases DOJ chose as examples, and one is about video: an agreement with Teachers Test Prep over complaints that its online video courses had no captions. For a business with a physical location, the background on who can bring a claim is in ADA website lawsuits, and the wider picture in what the ADA asks of a website.

This is general information, not legal advice. For your own situation, ask a lawyer.

Social video, embedded players and live streams

A YouTube or Vimeo video embedded on your page

The video is on your page, so it is part of your page for WCAG. Upload a finished caption file to the host rather than leaving its automatic track to stand in. YouTube's supported caption files include SRT and WebVTT, in their basic form without styling.

The player matters as much as the file. W3C's media player guide says some players are not accessible, and lists what one needs: keyboard support, a visible focus indicator and clear labels on its buttons. Tab to the play button and the captions button with no mouse; if you cannot reach them, the captions are there but out of reach. Our screen reader testing guide shows how to hear what the buttons are called.

One more rule catches background videos on home pages. Under 1.4.2 Audio Control (Level A), audio that plays by itself for more than 3 seconds needs a way to pause or stop it, or a volume control of its own. A muted loop is fine; a hero video that starts talking is not, unless the visitor can stop it at once.

Clips posted to social media

On a social platform, the platform owns the player, and some feeds give you no way to attach a caption file. Burn the captions into the clip itself before you post: open captions travel with the video wherever it is shared. Keep a closed-caption version for your own site, where the viewer can switch them off or enlarge them.

Live streams and webinars

1.2.4 was written for broadcast: W3C's Understanding note for 1.2.4 says it is not meant to require captions on two-way calls between people through web apps. A webinar you stream to the public is broadcast. W3C notes that live captions are usually written by professional real-time captioners or CART providers, and that a posted recording will probably need minor editing. YouTube's automatic live captions are English only and are switched on per stream, so they are a fallback, not a plan; the live options are compared in captioning services compared.

Video inside a slide deck

A video played from a presentation needs captions too. PowerPoint drops them when you use Compress Media or Optimize Media Compatibility, and when you save the deck as a video; how to add them and keep them is in accessible PowerPoint slides.

A caption quality checklist before you publish

Ten checks, in the order you meet them when you watch a video with the sound off. Tick each one as you go; the list prints on its own and copies as plain text for whoever edits the file.