In one paragraph
Video elements must have captions addresses the equivalent timed text needed for relevant audio in prerecorded synchronised media. Provide accurate captions that people can actually use, not merely a track element with a broken or incomplete file. Review exceptions and other media requirements separately. The supplied WebAuditOS report omits this rule's review outcomes and does not inspect every provider iframe, so no finding is not caption verification.
Why it matters
WCAG 1.2.2 requires captions for prerecorded audio content in synchronised media, except for clearly labelled media alternatives to text under the criterion's conditions. Captions include relevant speech, speaker information, and meaningful non-speech audio rather than dialogue words alone.
Captions need appropriate timing and readability as the video unfolds. A transcript can be valuable, but generally does not substitute for synchronised captions where the criterion applies. Descriptions of important visual information and live media requirements have separate scopes.
Open captions burned into the picture can be a valid delivery method when they provide the required equivalent information. A missing track is therefore not conclusive failure, while a declared track does not establish quality. Muting an available audio track by default does not automatically remove the need to provide its equivalent.
How WebAuditOS evaluates this finding
The supplied video-caption description looks at native video elements for a captions track. Presence satisfies its limited path, while absence can become human review because the engine cannot determine silent content, open captions, or equivalent information.
The source states that review results are omitted from the violations-only report, so the current integration raises no finding here. Provider iframe content is outside its native-element scope. A track present with missing files or incorrect text can remain undetected.
How to fix it
- Inventory prerecorded video, live streams, silent visual media, and provider embeds. Identify the actual speech and meaningful sounds rather than assuming every file has the same requirements.
- Produce equivalent accurate captions. Automatic recognition can draft them, but verify names, numbers, technical terms, speaker changes, sound cues, and synchronisation before publication.
- For a native player, a structural example is:
<video controls src="/media/product-demo.mp4">
<track kind="captions" src="/media/product-demo-en.vtt" srclang="en" label="English captions">
</video>Supply real matching resources and valid WebVTT. The example does not verify file retrieval, player controls, or caption quality.
- For provider embeds, configure captions through the supported platform workflow and test the actual published player. A top-level video-rule result cannot validate a video hidden inside another document.
- Confirm users can select and read captions using relevant input methods and devices. Test fullscreen, responsive presentation, language selection, and whether text obscures essential content.
- Keep captions aligned with edits to the recording and any translations. A valid language code and file extension do not prove the content matches the current audio.
- Review transcript, audio description, player names, keyboard operation, autoplay, and motion as separate needs. Meeting one media criterion does not grant complete audiovisual accessibility.
- Record a manual playback review and retain it after updates. Rescan other media and control checks, recognising that the supplied review-only caption rule cannot certify the task.
What the fix can break
Unreviewed automatic captions can materially change meaning. Assign an editorial owner rather than treating successful upload as accuracy validation.
A cross-origin file, player configuration, or access challenge can prevent captions loading even when source markup appears correct. Test actual public retrieval and playback.
Do not remove meaningful audio or mislabel a video as decorative solely to avoid producing an equivalent. Preserve the information and choose a usable accessible presentation.
Questions
Does a captions track prove compliance?
No. The file must load and provide accurate, synchronised equivalent information in a usable presentation.
Can burned-in captions be appropriate?
Yes, when they actually provide the required equivalent. A missing track alone does not decide the case.
Does a transcript always replace video captions?
No. Review the applicable media criterion; timed captions and transcripts serve different needs.
Which related checks help?
Review audio alternatives, autoplay sound, and the automation limits guide.
What severity does this get in the report?
None, in practice. Severity for an axe finding comes from the impact axe reports for the element that failed - this rule is declared critical, which would map to high - but in axe-core 4.13.0 the rule only ever returns pass or needs review, never a failure. Needs-review results are not included in our report, so this finding does not appear at all.