WCAG 1.2.7 — Extended Audio Description (Prerecorded)
When the natural pauses in a video aren't long enough to fit the necessary visual narration, the video must pause to let the description finish. Standard audio description with no room to breathe is incomplete description.
What this requires
For prerecorded synchronised media at Level AAA, extended audio description must be provided when the gaps between dialogue are too short to narrate the visual information. The player pauses the video, the description plays in full, and the video resumes. The result is a longer total runtime but a complete equivalent for blind users — no crucial visual is lost because there was no time to describe it.
How AI coding tools fail this
This is a production-and-tooling concern more than a code-generation concern. AI tools fail it by treating audio description as a single binary box: either there's a description track or there isn't. The nuance — "is the description complete, or did the producer cut content to fit it into existing pauses?" — never enters the conversation.
The code-level failure: custom video players that don't support
pausing the main track to let a description segment play. Native
<track kind="descriptions"> in browsers that support it can handle
this; custom players written from scratch generally don't.
Edge cases
- Production overhead. Extended description usually means re-editing the source video, which is expensive. Plan for it on flagship content, not on every clip.
- Alternative: a full text transcript covering all visual information satisfies 1.2.8 at AAA and is sometimes easier to produce than a separate extended-description video.
- The default video is still the default. The extended version is in addition to, not in place of, the standard one.
- Browser implementation of
kind="descriptions"has historically been weak. A separately produced video file is the most reliable delivery method.
How Jeikin handles this
This is a manual review item. The scanner records each <video> and
asks the reviewer whether the standard audio description is complete
or whether extended description (or a full transcript covering
visuals) is needed. The dashboard keeps the answer as evidence.