Rampsight

All rules

Audio and video

Audio needs captions or a transcript

Why it matters

Recorded speech is inaccessible to anyone deaf or hard of hearing, and to everyone in a noisy place or without headphones. Nothing on the page conveys what was said.

How to fix it

  1. Add a <track kind="captions"> file to the <audio> element.
  2. Publish a text transcript next to the player as well — it is also what a search engine reads.
  3. Include the speaker names and any meaningful non-speech sound.

In markup

Fails
<audio controls src="interview.mp3"></audio>
Passes
<audio controls src="interview.mp3">
  <track kind="captions" src="interview.vtt" srclang="en" label="English">
</audio>

What the check cannot tell you

The check sees whether a captions track exists, not whether it is accurate or synchronised. A wrong transcript passes.

The requirement behind it

This rule is one way of failing the following WCAG 2.2 success criteria. The W3C, which writes the standard, explains each of them:

Further reading

Deque University on audio-caption (opens in new tab)

Deque maintains axe-core, the open-source engine our scanner runs. Their page documents the rule as the engine implements it.

Automated checks find failures a browser can prove. They cannot judge whether a caption is accurate, whether an alt text is useful, or whether a page can be operated end to end with a keyboard — those need a person.