• Success Criterion 1.2.2
  • Conformance level A

YouTube Auto-Captions Never Corrected for Accuracy

1.2.2 — Captions (Prerecorded)

Scenario

Setting

A fitness app's workout tracker

What’s wrong

Auto-captions (YouTube auto-generated (speech recognition)) left uncorrected: wrong words, no punctuation, no speaker IDs — fails accuracy/equivalence.

Example

<iframe src="https://www.youtube.com/embed/abc123?cc_load_policy=1" allowfullscreen></iframe>
<!-- Auto-captions read: "now grab your cattle bell and do ten bur peas
     then switch to the lung's watch your form" -->

Why it matters

A deaf user trying to follow the workout reads "cattle bell" and "bur peas" and has no idea these mean kettlebell swings and burpees, so they can't follow the routine.

How to test

Turn on YouTube/platform auto-captions and compare 30 seconds against the actual audio: uncorrected wrong words, missing punctuation, and no speaker labels mean it fails.

How to fix

Auto-generated captions are a starting draft only; someone has to review and correct terminology before publishing fitness or technical content.

<iframe src="https://www.youtube.com/embed/abc123?cc_load_policy=1" allowfullscreen></iframe>
<!-- Corrected caption track uploaded through YouTube Studio, replacing
     the auto-generated one: "Now grab your kettlebell and do ten
     burpees, then switch to lunges. Watch your form." -->

Outcome

A deaf user follows the corrected captions and completes the full kettlebell circuit without guessing at exercise names.

Who is affected

Deaf and hard-of-hearing users following captioned workouts get garbled exercise names that don't match any move they recognize.

Learn more