Imagine that you are reviewing a training, and when you select "Next", a video starts playing automatically. You frantically start looking for a pause or stop button but find none. Now, imagine the same situation but with someone using a screen reader, trying to find a pause or stop button while the screen reader and video sound play simultaneously, interfering with each other.
WCAG 1.4.2: Audio Control (A) states that the user should be able to stop or pause any audio that plays automatically for more than 3 seconds. One easy way to comply is by providing a pause or stop button for any multimedia with sound that plays automatically for more than 3 seconds. However, it's generally considered a better solution to allow users to start the multimedia containing audio after landing on the content page instead of giving them the option to stop it. So it's best to not have multimedia with audio auto-start but start at the user's control.
Talking about audio content, there is an advanced criterion that relates to the relationship between foreground and background audio within a track.
WCAG 1.4.7: Low or No Background Audio (AAA) states that any pre-recorded audio that contains speech in the foreground should not contain background audio or users should be able to turn off the background audio. Any background audio, if present, should be 20 dB lower than the foreground speech audio.