Being able to add to the song with commands like 'add a psytrance bass line', even within predefined parameters, to dynamically generate an entirely new bass line from other songs in the genre.
Maybe you could instantly add an improvised violin melody by telling it a style given that the chords/key from the human band are consistent.
Sentiment analysers could tweak the music based on crowd reaction towards musician defined goals and learn those pre-sets over time.
If music is a synchronisation layer between humans, maybe machine learning could help us to synchronize even more closely.
Being able to add to the song with commands like 'add a psytrance bass line', even within predefined parameters, to dynamically generate an entirely new bass line from other songs in the genre.
Maybe you could instantly add an improvised violin melody by telling it a style given that the chords/key from the human band are consistent.
Sentiment analysers could tweak the music based on crowd reaction towards musician defined goals and learn those pre-sets over time.
If music is a synchronisation layer between humans, maybe machine learning could help us to synchronize even more closely.