WebsightWhat it asks for, in plain English
If you stream video with sound live, such as a webinar, a town hall, a product launch or a church service, the speech needs captions as it happens, not afterwards.
Live captions are usually produced by a trained captioner typing in real time, or by automatic speech recognition that is monitored and corrected. Unchecked automatic captions are a weak fit for names, jargon and multiple speakers, so test them with your actual content before relying on them.
This criterion is about video with sound. A live audio-only stream, such as a radio show, is not required to have captions at Level AA, though a transcript afterwards is good practice.
This is an independent explanation. The binding wording is the W3C Recommendation, and the W3C's own Understanding 1.2.4 page has the full intent, techniques and definitions.
Who it affects
A deaf attendee at your online all-hands can take part in the questions session only if the captions keep up with the speakers.
Someone joining from a noisy site office reads the captions because the audio is drowned out by machinery.
A person with an auditory processing difference uses captions to pin down fast or overlapping speech.
How to check it
- List every live video event you run or embed, including recurring ones such as weekly webinars.
- Join a live session as an attendee and turn captions on. Check the control is easy to find before the event starts.
- Watch the captions for a few minutes against the speech. Note the lag, missed names and how speaker changes are shown.
- Check the caption text is readable at the size and position the player uses, and can be resized.
What Websight checks for 1.2.4
Websight has no automated check for this criterion. It lists it in the manual checklist that comes with every report, so it is not forgotten.
Only a person can check this. A scanner cannot join a live stream, and captions are part of the streaming platform rather than the page markup.
Practical tips
- Send the captioner (human or automated) the names, product terms and agenda in advance; accuracy jumps.
- Choose a streaming platform with captions built in rather than bolting a separate service on.
- Post a corrected transcript with the recording afterwards; it is also what 1.2.2 needs once the video is on demand.