Media accessibility check

The Digital Storytelling Project is graded in part on accessibility, and the requirements are not ours. This page asks eleven questions about a video and returns the actions the standards require, in the order the standards put them, beginning with WCAG Level A, then Level AA, then the federal caption and player guidance, then the production and consent rules this course applies, and last the imagery questions. Each action names the standard or the agency it comes from. Nothing is saved or sent anywhere.

The Week 11 lab is a caption sprint, and the second half of this page is the arithmetic from that sprint: a word error rate from your own count on your own footage.

The questions need JavaScript. The requirements themselves are readable without it, at W3C on captions, Section508.gov on synchronized media, The A11Y Project checklist, and the Overlay Fact Sheet.

Captions

Description and transcript

On-screen text and player

Consent and production lane

Imagery

Required actions

A question left blank is reported as unanswered and marked in place, because an unchecked requirement cannot be counted as met.

Caption error count

The standard for captions is verbatim speech, so the only acceptable error count is zero, and counting establishes how far a generated track is from that. Count on a sample of your own footage: take two minutes of it, count the words in the audio, then count the errors in the generated caption track against what you hear. A wrong word, a missing word, an inserted word, and a wrong speaker label each count as one error.

Words in the sample

Errors you found

The comparison worth having is with the published benchmark. An independent test of eleven commercial speech-recognition services on real lecture recordings found error rates varying too much across services and recordings to trust any single vendor number, which is why the count that decides your correction pass is the one you take on your own audio.

Sources

W3C Web Accessibility Initiative, captions. Synchronized captions for prerecorded video are WCAG Level A, audio description of visual information the audio does not convey is Level AA, and automatically generated captions do not meet the requirement unless they are confirmed to be fully accurate.
Section508.gov, synchronized media. Federal guidance on captions, audio description, and transcripts, including the requirement that caption and audio-description controls be at the same menu level as the volume control.
FCC 14-12, closed captioning quality, in the Federal Register. The four caption quality standards, accuracy, synchronicity, program completeness, and placement, effective January 15, 2015. The codified rule is 47 CFR 79.1.
The A11Y Project checklist, with each item tied to a WCAG success criterion. Contrast thresholds are testable at the WebAIM Contrast Checker.
Overlay Fact Sheet, signed by more than 1,000 accessibility practitioners including specification authors. Relevant the moment anyone offers a widget in place of a fix.
Stella Young on inspiration porn and SAIH's social-media guide, the two sources behind the imagery questions.
Automatic speech recognition in higher education, the eleven-service benchmark on real lecture recordings.