AI in digital sound management/processing/streaming?

How about implementing AI in hi-fi equipment and digital sound management/processing/streaming? Do you have any hints?

For audio? There’s some interesting things, source separation is getting quite good (it’s been around a long time, hell Abbey Road’s guys were messing around with this for ages to do The Beatles stuff) . Not really a fan of ML guided DSP yet.

For music metadata, some of the ML entity resolution stuff is interesting, genre classification isn’t really ‘good’ yet, it’s too broad IMO