alt Hacker NewsYou could say I'm a music snob, big time. I can't stand any of the streaming services, because they only have a small fraction of my favorites (which is variations of discord jazz, often in other genres. I like when music decomposes into noise and then restructures again.) Due to my interest there, I've done a deep hole with AI music generation, not the services, but developing the models and exploring the open weight models being released.
There will be quality real art music created by these systems, but not by those that prompt alone. This is a whole new level of instrument, and the levels of control beneath are there to seriously transform one's thoughts to music, and melody, and that composed symphony of separate elements into a symphony of intended meaning.
Perhaps traditional music and this form of music should be treated separate. The distinction between AI music that is prompt-only and what can be created from a deeper set of controls is immense, and is not distinguished at this time, and may never be with how surface level this entire public assessment of AI music happens to be.
My take is that it will be somewhat similar though perhaps more complex to using sample based libraries, whether it’s for electronic music, symphonic loops, etc., I feel like this will just be the next generation of that kind of music making. I find it very unlikely that it will replace the kind of music making that comes from playing actual instruments and recording them from a live human player but it could still be interesting in its own right.