WCAG 1.2.2 Captions (Prerecorded)

What it asks

Captions are text shown on or next to the video, in sync with it, carrying the dialogue and the sounds a viewer needs to follow what is going on: who is speaking, a phone ringing, music that sets the scene. They are not the same as a transcript. A transcript is read separately; captions are read while watching.

This rule covers all prerecorded video that has an audio track: marketing videos, tutorials, recorded webinars, lectures, interviews and adverts. The captions can be closed, where the viewer switches them on, or open, where they are part of the picture. Either way they have to exist, and they have to be right.

There is one exception. A video that is itself an alternative for text on the page, and is clearly labelled as such, does not need captions, because the text is the alternative.

Captions have to be accurate. Auto-generated captions published without a review pass are the most common way this criterion ends up as Partially Supports in a conformance report: the captions are there, but they say the wrong words, miss the names and lose the meaning. Live video is covered separately by 1.2.4 Captions (Live).

Its official name is success criterion 1.2.2 Captions (Prerecorded), at Level A. It was added in WCAG 2.2, under guideline 1.2 Time-based Media and the principle Perceivable. The W3C’s normative text says, in part:

“Captions are provided for all prerecorded audio content in synchronized media, except when the media is a media alternative for text and is clearly labeled as such.”

From WCAG 2.2, W3C. Full explanation: Understanding 1.2.2 Captions (Prerecorded).

Who it helps

Deaf and hard of hearing people, for whom captions are how they follow a video at all. People with auditory processing difficulties, who hear the sound but struggle to make it out.

It also helps people watching without sound in a shared space, people whose first language is not the one spoken, and people learning the subject, who read along as they listen.

What passes and what fails

These pass

• A video element with a track kind="captions" file that was checked by a person against the recording.
• A YouTube embed whose owner uploaded or corrected a caption file, rather than leaving the automatic ones.
• A video with open captions burned into the picture, as long as they are accurate and readable.
• A video that reads out an article word for word, labelled "Audio version of this article", next to the article.

These fail

• A product video with no captions at all.
• Automatic captions published without anyone reading them, with names and technical terms garbled.
• Captions on the main marketing video but not on the tutorials in the help centre.
• A transcript below the video offered in place of captions. A transcript meets 1.2.1 and can meet 1.2.3, but it is not synchronised.
• Captions that carry the dialogue but none of the sounds, such as a doorbell the story depends on.

How to test it by hand

1. Find every video on the page with a soundtrack, including embedded players from video services.
2. Play each one and look for a captions control. Switch captions on.
3. Watch with the sound off. Can you follow everything that is said and every sound that matters?
4. Watch again with the sound on and compare. Note wrong words, missing names, lost speaker changes and timing that lags behind.
5. Where captions are burned into the picture, check they are readable at the size the video is shown.
6. Note every video with no captions, and every video whose captions are incomplete or inaccurate.

What automated testing can decide

What a machine can decide

Automated testing can find each video on the page and see whether a caption track is declared, either as a track element or through a player that exposes its captions. A recorded video with sound and no captions anywhere is a clear failure, and that is a result automated testing can report.

What it cannot

A caption track proves there are captions, not that they are good. Whether they match the audio, whether they are in time, and whether they carry the sounds that matter all need someone to watch the video with them on. A third-party player often keeps its captions out of reach of any test of the page around it, and captions burned into the picture are invisible to code.

How Dreven handles it

Dreven finds the recorded videos with sound on your pages and shows you each one with the captions it found, the page and the player named. A reviewer watches the video with captions on and decides each result.

The report gives one of four results for each video, passed, failed, not applicable or can’t tell, and never counts a video it could not see into as a pass.

No automated check makes a site conform. In Dreven, a person reviews the results and signs the report.

Related criteria

• 1.2.4 Captions (Live) (Level AA). Live video needs captions as it happens.
• 1.2.3 Audio Description or Media Alternative (Prerecorded) (Level A). The same video also needs a spoken description or a text alternative for what is shown.
• 1.2.1 Audio-only and Video-only (Prerecorded) (Level A). Recordings with only sound need a transcript instead.
• 1.4.2 Audio Control (Level A). Video that plays sound on its own must be easy to pause or mute.

Questions people ask

All WCAG 2.2 criteria, in plain words