Captions, transcripts and audio description for video
Video is everywhere on modern websites: hero loops, product demos, recorded events. People who are deaf or hard of hearing need captions; people who are blind need to know what's happening on screen.
What WCAG asks for
- Prerecorded audio only (like a podcast), Level A: a text transcript (1.2.1).
- Prerecorded video only, no sound, Level A: a transcript or an audio track describing what's shown (1.2.1).
- Prerecorded video with sound, Level A: captions (1.2.2), plus a transcript or audio description (1.2.3).
- Prerecorded video with sound, Level AA: audio description of important visual information (1.2.5).
- Live video with sound, Level AA: captions (1.2.4).
- Anything that moves for more than five seconds and starts automatically, Level A: a way to pause, stop or hide it (2.2.2). That includes silent background video.
Captions, subtitles, transcripts: the difference
- Captions are synchronized text of everything that's said, plus important sounds: “[phone rings]”, “[laughter]”, speaker names when it isn't obvious.
- Subtitles usually translate speech for people who can hear but don't speak the language. They often leave out sounds, so they don't count as captions on their own.
- Transcripts are the whole thing as text on the page: speech, sounds, and a description of what's shown. They help deafblind users with braille displays, and anyone who'd rather read.
- Audio description narrates important visual information that the soundtrack doesn't cover: on-screen text, actions, scene changes.
How to make captions
- Start from an automatic transcript: most video platforms and editing tools generate one.
- Edit it by hand. Automatic captions get names, numbers and menu items wrong, and they leave out punctuation that carries meaning.
- Add speaker names and important sounds.
- Export it as a caption file (WebVTT .vtt is the web standard) and attach it to the video, or upload it to your video host.
- Watch the video with captions on, from start to finish, to check the timing.
<video controls>
<source src="/media/how-we-cook.mp4" type="video/mp4">
<track kind="captions" src="/media/how-we-cook.en.vtt" srclang="en" label="English" default>
</video>Audio description without re-editing
The easiest way to meet audio description is to plan for it: when you script a video, say out loud what's on screen. “Here's our new patio, open from noon” describes itself. For existing videos, a separate described version, or a transcript that includes the visual details, can help.
Background and hero video
A silent looping video in a page header has no speech to caption, but it still needs a visible pause button if it plays for more than five seconds, and it should respect the visitor's “reduce motion” setting. Mark it as decorative if it carries no information.
Practical takeaways
- Never publish automatic captions without reviewing them.
- Put a transcript on the same page as the video, or link to it right next to it.
- Give every autoplaying video a pause control, and don't autoplay sound.
- Use the video player's built-in caption support rather than burning captions into the picture, so people can resize them.
What Scanpoof checks, and what needs a person
Found automatically
- Player buttons with no accessible name
- Embedded video frames with no title
Needs a person
- Whether captions are accurate and in sync
- Whether important visuals are described
- Whether captions exist at all, including on videos embedded from other platforms
- Whether a transcript is complete
For some of these, Scanpoof's AI writes a fix, tests it on your live page and waits for you to approve it. How AI Fix works.