I a bit bothered by the usage requirement of the audio element in MO. The text says:

Usage
A REQUIRED child of the par element unless its sibling text element refers to audio or video media, or to textual content intended for rendering via text-to-speech, in which case it is OPTIONAL (see 7.3.2.4 Embedded Media).

As read, this sounds like an untestable requirement to me, unless it is tested run-time. After all, how does a checker know that an HTML file, linked through a text element, is used with text-to-speech? We do not define normatively anything for tts after all.

Maybe EPUBCheck found a trick that I miss (@rdeltour ?). Otherwise, it may be cleaner to replace that requirement by something like:

An OPTIONAL child of the par element. If the element is missing, its adjacent text element either refers to audio or video, or is supposed to refer to a textual content intended for rendering via text-to-speech.

I think that content-wise we are saying the same thing, but the checker is not under the obligation of checking this.