What if they did HDR for audio? So an audio file can tell your speakers to output at 300% of the normal max volume, even more than what compression can do.

Isn't that just by having generally low volume levels? I'm being pedantic, but audio already supports a kind of HDR like that. That said, I wonder if the "volume normalisation" tech that definitely Spotify, presumably other media apps / players / etc have, can be abused to think a song is really quiet.