About time
For some time I've thought it strange that we are forced to take the mix that a sound engineer deems appropriate. With the flexibility of file formats it should be possible to deliver sounds as discrete lumps, using as many voices as necessary and combined with metadata to describe where that sound emanates from.
If you've got the master tapes then pretty much any recording could be re-encoded accordingly.
Imagine listening to a pop group, but you'd rather Ringo sounded like he was playing in the toilet down the hall - entirely possible if each master mix channel was delivered separately. Want to auto-tune Paul's latest efforts? No problem. Fancy Debbie H on your lap? Sorry, I digress too much...
As soon as XBMC can decode Atmos I'm in (as long as I can hide the extra speakers behind the sofa).