Multimedia content must have caption support for audio. Captions must be accurate, must include dialogue, the individual speaking, and any relevant audio information.
Siteimprove helps for discovering video content (QA > Inventory)
Test procedure
Video content can be found with the assistance of the enterprise Siteimprove app. From the Dashboard of the site you are assessing, navigate to QA in the navigation, then the Inventory option. Click on the "Video" link to get a list of video content on the site. Please note that this includes multimedia video (the subject of this criterion) and silent video.
Captions must be present.
Closed or Open captions are both valid.
The captions MUST be accurate to the audio.
Commonly, YouTube automated captioning is inaccurate, and rarely uses punctuation. These will more often than not fail to be sufficient.
The captions MUST include all relevant non-verbal noises.
The captions MUST denote a change of speaker.
It's strongly recommended to always denote speaker changes in captions, to support users with cognitive disabilities, but if the speaker change happens on-screen (as opposed to background audio/narrator voices), it's a warning not a failure.