I know “HDR” is the hot phrase these days, but what I really want is LOW dynamic range audio. Ideally authors should be able to supply two audio tracks: one with full dynamic range and one with restricted dynamic range, and if the restricted one is absent then YouTube should auto-generate it on the server (not client!) side. It’s a little hard to do this well in real-time on low-end consumer hardware but not too difficult to automate without time constraints.
Aside from, say, classical music, there is nothing I listen to that I want to have significant changes in volume. I don’t need explosions to wake the neighbors, thanks.