Audio description for video: what it is and when WCAG requires it

Audio description lets people who cannot see a video follow it by ear. WCAG's glossary also calls it video description and descriptive narration. A describer, or the speaker, says what the picture shows where the sound leaves a gap: who has entered the scene, what a chart shows, what a title says. W3C's page on description of visual information says description serves people who are blind and others who cannot see the video adequately, and its Understanding page for success criterion 1.2.5 adds people with cognitive limitations who find it hard to interpret what is happening visually.

In standard description the narration goes into pauses in the existing dialogue, so it does not talk over speech. Captions do a different job: they give people who cannot hear the sound a text version of it. See video captions and transcripts for that side and captions, alt text and figcaption for the image sense of the word. For still images, alt text plays the part that description plays for video.

This guide covers what WCAG asks at each level, when no description is needed, the ways to provide it and how to check a page. It is general information about web accessibility standards, not legal advice.

What WCAG requires for audio description

Five success criteria in WCAG guideline 1.2 deal with description of video or a text version of it. All exist in WCAG 2.0, 2.1 and 2.2: W3C's list of what WCAG 2.1 added includes nothing from guideline 1.2, and our page on what is new in WCAG 2.2 lists nine additions, none in 1.2.

Sources: W3C's Understanding pages for 1.2.1, 1.2.3, 1.2.5 and 1.2.7, and WAI's transcripts page for 1.2.8.

How 1.2.3 and 1.2.5 fit together

W3C's Understanding page for 1.2.3 says 1.2.3, 1.2.5 and 1.2.8 overlap somewhat, to give authors some choice at the minimum conformance level and to add requirements at higher levels. At Level A you choose between audio description and a full text alternative. At Level AA you must provide audio description: if you chose description at Level A you have already met it, and if you chose text, description is an extra requirement. A text alternative alone therefore meets Level A only. AA is the level the laws we track name when they name one (see WCAG conformance levels), so it is the level to plan for.

A full text alternative is not a plain transcript. W3C says it reads like a screenplay: dialogue in order, with descriptions of the visuals and of non-speech sounds. A transcript of speech alone does not meet it.

When a video does not need audio description

The requirement follows the content. W3C WAI's test is one question: does the video show visual information needed to understand what it communicates? If not, no description is needed. The usual cases:

• The soundtrack already says it. W3C notes under 1.2.3, 1.2.5 and 1.2.7 that no description is necessary when the audio already carries all the important information in the picture. For example, a screen recording whose narrator says everything important about what is selected and what appears can fall in this group.
• A single speaker. W3C's technique G203 covers one speaker against an unchanging background: a static text alternative giving the setting, credits and on-screen text such as the speaker's name is enough. It does not apply when several speakers are identified only by on-screen text. Then audio description is the answer.
• Audio-only content. A podcast has no visual information, so WAI says description is not needed, though it still needs a transcript at Level A. A recording published as video is video, and the video criteria apply if the picture carries information.
• Live video. Criterion 1.2.5 covers prerecorded video. WAI says live description meets a user need but is not required to meet WCAG. In our reading, a live event posted afterward is prerecorded.
• Media that only repeats text. W3C's Understanding pages for 1.2.3 and 1.2.5 leave out a media alternative for text: media that presents no more information than is already presented in text and is clearly labeled as such.

Silent video is not exempt: criterion 1.2.1 asks for a text alternative or an audio track with the same information.

What to describe, and what to leave out

W3C's Understanding page says description covers actions, characters, scene changes and on-screen text that are important and not already described or spoken in the soundtrack. WAI's description guidance adds tips:

• Describe what a viewer needs to understand the video, not every detail.
• Describe objectively, without comment, in the present tense, active voice and third person.
• Cover on-screen text, such as an opening title, a web address shown at the end or a speaker's name.
• Time the description to the visual or right before it.
• Use a neutral voice that is easy to tell from the video's speakers, and lower the main audio while it plays.

Pauses matter too. W3C's failure F113 says 1.2.5 fails when important visual information goes undescribed although usable pauses in dialogue exist. It notes that a pause can be unusable if it is very short (it gives less than 2 seconds as an example) or far from the visual.

Four ways to provide audio description

WAI says the right method depends on the video, the space in its audio and what the media player supports. The table follows its description page.

W3C's order of questions is a good default. For a new video whose speakers can describe the visuals as they go, integrate the description. Otherwise, if your player reads description from a text file, use a timed text file. If it supports a separate audio track and the main audio has room, record one. If none of that fits, make a described version. W3C says the first options are usually best, though you may choose another. Integration is easiest when planned early: WAI's guidance on creating audio and video content says planning for description in advance is fairly simple for many videos, and suggests saying what a chart shows in place of "as you can see on this chart".

If the pauses are too short

Extended audio description pauses the video so the narration can say more than the gaps allow (criterion 1.2.7, Level AAA). W3C says it is used only when the sense of the video would otherwise be lost. WAI does not suggest building in functionality that pauses the video for description in most cases, because it requires extra coding and gives a less-than-optimum user experience. It suggests recutting scenes longer where the source video allows, or recording extra time in a scene when you plan the video, and otherwise leaving a static image in the video while the description plays. W3C's technique G8 provides extended description as a second version of the video, in which the picture halts while the narration plays.

The track element, and why not to rely on it alone

HTML can declare a description track. This example is adapted from W3C's technique H96:

<video controls>
  <source src="demo.mp4" type="video/mp4">
  <track kind="descriptions" src="demo-descriptions.vtt" srclang="en" label="English descriptions">
</video>

W3C lists H96 only as an advisory technique and noted in September 2023 that browsers had no native support for it, although JavaScript polyfills existed. WAI's media players page says that, to the best of its knowledge, AblePlayer and video.js support description from a text file when there is space in the audio. Check what your player and video host support before you choose a method.

Does a law require audio description?

Web accessibility rules usually name a WCAG version and level, so what they require for video follows from that choice. Audio description is in scope wherever the named level is AA, unless the rule carves it out. Three examples, with the caution that which rule and exceptions apply to you is a question for a lawyer:

• US state and local governments. The Department of Justice's ADA Title II web rule sets WCAG 2.1 Level AA, with compliance dates of April 26, 2027 for populations of 50,000 or more and April 26, 2028 for smaller ones and special districts. The page does not name 1.2.5. In our reading it is part of that standard, because it is a Level AA criterion, so prerecorded video in scope needs description unless an exception applies. See the Title II deadline guide.
• Ontario, Canada. The Ontario government's AODA website guidance requires WCAG 2.0 Level AA except 1.2.4 (live captions) and 1.2.5 (pre-recorded audio descriptions), and its list of Level A criteria still includes 1.2.3. See accessibility law in Canada.
• The EU. EN 301 549 v3.2.1, the current reference standard, is aligned to WCAG 2.1 Level AA, which includes 1.2.5. See EN 301 549 and European Accessibility Act compliance.

This guide covers web content under WCAG only, not any separate duty a rule may place on a particular kind of video service. Our which accessibility laws apply to me questionnaire shows the sources for each law it lists.

Check it yourself

1. List the videos. Note every prerecorded video on your key pages, embedded ones included, and which have sound.
2. Ask W3C's question. Does it show anything needed to understand it that the sound does not say? If not, write down why.
3. Find the description. Look for a described version linked next to the video, a switchable audio track or a description track your player reads aloud. A full text alternative meets Level A only.
4. Listen without looking. Our quick test: look away, switch the description on and play the video. Can you follow what happens? Is on-screen text covered? Does the narration land at or just before the visual?
5. Test the control. W3C's test for G78 is that a way to turn description on exists, in the content, the player or the operating system. WAI's media players page says players need to work from the keyboard, so try it without a mouse.
6. Record the result for each video as pass, fail or not applicable.

Tools help with the record-keeping, not the judgment. The WCAG checklist tracks all 55 Level A and AA criteria as pass, fail or not applicable, with notes and a CSV export, so 1.2.1, 1.2.3 and 1.2.5 sit beside the rest. The accessibility bookmarklet checks the rendered page for things like missing names and low contrast, and has no check for video description. Scan my site for free flags autoplaying audio or video (autoplay and motion have rules of their own, covered in animation and reduced motion), but nothing in its checks can tell whether a video needs description or has it. Our WCAG criteria reference lists 1.2.5 as a manual check for that reason, and what we check sets out the limits.