Do we need captions for every video?

For every recorded video with sound where the audio information is needed to understand it. A silent mood video doesn't need captions — but it does need a text alternative if it shows information.

Are the video platform's automatic captions enough?

As a starting point, yes; as the finished result, no. They need to be reviewed and corrected. What matters is whether the caption reflects what's said — not whether a track exists.

What's the difference between audio description and transcript?

An audio description describes what's on screen during pauses in speech. A transcript is the complete text of picture and sound to read afterwards. WCAG allows either for recorded videos — one of the two must be present.

This translation has not been editorially reviewed yet. The German version is authoritative. Deutsche Fassung →

Captions and audio description: what's required for video and audio

Recorded videos with sound need captions, recorded videos with visual information additionally need an audio description or an equivalent text alternative. Live broadcasts with sound need live captions. Audio-only or video-only recordings need a text alternative.

The most common mistake is treating auto-generated captions as good enough. They're a good starting point, but unreviewed, they contain errors in proper names, technical terms, and numbers — exactly the words that matter most. A caption that states the wrong price is worse than none at all.
The second most common mistake concerns silent videos running as background. They don't need captions — but if they show information that's not stated anywhere else, that information needs a text form.
This guideline is the area where automated checking can do the least: that a caption track exists can be established; whether it's in sync and content-accurate cannot.

Deutsche Fassung — mit allen Prüfpunkten dieses Themas