WCAG 2.2 · Guideline 1.2 · Perceivable
Time-based Media
Audio and video need captions, transcripts, and description, because sound and picture do not reach everyone.
A video plays for everyone and what it says does not reach everyone. Deaf and hard-of-hearing people need captions, blind people need the visuals described, and deafblind people need a transcript they can read in braille. This guideline covers all three, live and recorded. What it asks of you depends on what your media actually holds, because every criterion here carries its media type inside its own name. A podcast answers the audio-only rules and nothing else. A silent demo video needs no captions at any level.
Five A/AA success criteria live here, out of nine. They cover captions for recorded and live video with sound, a transcript or an alternative for audio-only and video-only recordings, and description of what happens on screen while nobody is talking about it. The four above them, at Level AAA, add sign language and extended description. Guideline 1.2 is not a thing you pass or fail on its own, though. An audit runs against those five, and which of the five reach you is decided by the media you publish rather than by the guideline.
Test rules for this guideline
The W3C files these ACT rules against guideline 1.2 as a whole rather than any single criterion under it.
- Audio element content has transcriptA tool finds candidates, you decide
- Audio element content is media alternative for textA tool finds candidates, you decide
- Video element auditory content has captionsA tool finds candidates, you decide
- Video element visual-only content is media alternative for textA tool finds candidates, you decide
- Video element visual-only content has transcriptA tool finds candidates, you decide
- Video element visual-only content has audio track alternativeA tool finds candidates, you decide
- Video element content is media alternative for textA tool finds candidates, you decide
- Video element visual content has audio descriptionA tool finds candidates, you decide
- Rules
- 5
- Level A
- 3
- Level AA
- 2
- Human testing only
- 4
All 5 Time-based Media rules
- 1.2.1Level AAudio-only and Video-only (Prerecorded)A recording with only sound needs a text alternative carrying the same information.Human testing6 fixes
- 1.2.2Level ACaptions (Prerecorded)Every prerecorded video with sound needs synchronized captions.Partly scannable8 fixes
- 1.2.3Level AAudio Description or Media Alternative (Prerecorded)Where a video shows something that matters and never says it out loud, WCAG 1.2.3 gives you a choice.Human testing13 fixes
- 1.2.4Level AACaptions (Live)A live event carrying sound alongside pictures or slides needs captions while it is happening.Human testing5 fixes
- 1.2.5Level AAAudio Description (Prerecorded)At Level AA the text alternative stops being an option.Human testing11 fixes
Level AAA in this guideline
4 enhanced criteria sit under 1.2. They are outside the level almost every law names, so our audits treat them as reference rather than scope. Some are still worth adopting, and the Level AAA hub says which.
- 1.2.6Level AAASign Language (Prerecorded)Prerecorded audio that runs synchronized with something else needs a sign language interpretation, and that sits on top of captions rather than replacing them.Not graded4 fixes
- 1.2.7Level AAAExtended Audio Description (Prerecorded)Some videos leave no room for description.Not graded4 fixes
- 1.2.8Level AAAMedia Alternative (Prerecorded)A media alternative is a text document carrying everything the video says and everything it shows, in the order it happens.Not graded5 fixes
- 1.2.9Level AAAAudio-only (Live)Live audio with no picture needs a text alternative carrying the same information, available while the audio is running.Not graded3 fixes
What goes wrong here
These are the failures we find repeatedly under 1.2, across sites of every size.
- Machine captions go out unread. Accuracy is the requirement, so an automatic pass is a first draft. The failure mode is quiet. The machine drops the word "not", reverses a safety instruction, and never once looks wrong.
- Captions carry the dialogue and drop everything else. Speaker changes and meaningful sound go missing, and the reader gets the words without the scene they happened in.
- Visual information that the soundtrack never states goes undescribed. If a chart appears on screen in silence, somebody listening has no idea it was there.
- A live stream with sound runs without real-time captions. Nothing can be fixed afterwards, which makes this the hardest requirement in the group to retrofit.
- The player itself cannot be operated, or starts loud on its own. Those are keyboard and audio-control failures, and they sit outside this guideline while breaking the same video.
Who it affects
- Deaf and hard-of-hearing people, for whom captions are the content rather than a convenience. Meaningful non-speech sound is part of that content, not a nice extra.
- Blind people, who need description when something happens on screen in silence. Where the narration already says it, no extra description is owed.
- Deafblind people reading through braille, who need a text alternative rather than either channel. This is the audience a transcript genuinely rescues.
- People who take in written information more reliably than spoken, which is a wider group than deafness alone.
- Audiences reading a translated track. Subtitles translate; captions carry the sound. The two overlap where a translated track also carries speaker changes and non-speech information, and where it does not, translation is not an accessibility deliverable.
How to work through it
- 1Inventory every piece of audio and video, then mark each one prerecorded or live, and with sound or without. Those two answers decide which criteria apply, and getting them wrong is how teams audit the wrong rule.
- 2Play each one with the sound off. Write down what you lost. That list is what the captions have to carry.
- 3Play it again with the screen off. What you lose this time is what description has to carry, and if the soundtrack already said it, you owe nothing further.
- 4For prerecorded video with sound, check the A and AA obligation properly: captions at A, and at AA either audio description or a full text alternative for the visuals. A separate transcript is a real kindness and it is not what these two levels ask for.
- 5Read any transcript you do publish on its own, with the media closed. It has to stand up alone, including the outcomes of anything interactive inside the video, or it is not the alternative it claims to be.
How the levels build
Which criteria apply depends entirely on what kind of media you have, so read this as a routing table rather than a ladder. Level A wants a text alternative for audio-only recordings and a text or audio alternative for silent video. It also wants captions on prerecorded video with a soundtrack, plus either description or a full text alternative for visuals the soundtrack never states. Silent video needs no captions, because the captions rule is scoped to media where sound and picture run together. There is also an exception worth knowing at every level: where the media is itself a clearly labelled alternative to text already on the page, it is not asked to repeat the work. Level AA adds live captions for streamed events with sound, and audio description for prerecorded video. Level AAA is four more criteria, not two. Sign language interpretation, extended audio description for when the natural pauses are too short, a full media alternative for prerecorded synchronized media, and text for live audio-only. That last one is why a live podcast without a real-time transcript is an AAA gap rather than an AA failure.
Other Perceivable guidelines
- 1.1Text AlternativesAnything that is not text needs words attached, because text can be read, spoken, or felt.
- 1.3AdaptableStructure has to live in the code, not only in the way the page looks.
- 1.4DistinguishableContent has to be easy to see and hear as it ships, with no adjustment needed.
Part of the Perceivable principle · browse by level: Level A · Level AA · or the full 55-rule library.
See how your site does against these rules.
An expert review plus a real blind screen-reader user, on up to 10 pages, every finding with its screenshot, criterion, and fix. $499, report in 5 business days.