Raw LLM Responses

If you're a real musician, producer, or lyricist, you have little to fear from AI music. You can do vastly more with it (assuming you want to) than very nearly everyone who's using such services. Nearly 100% of people using AI to generate music have no ear for what it creates. They have never made any music themselves and therefore don't know how to understand its output. They don't know the difference between what sounds good to them, and what sounds good to anybody else. They have no concept of "the guitars sound muddy." I tried to explain to one of them that one of the services' voice generator sounds like a vocoder. I could not explain it to him at all. (Look up anything made with Udio that has a baritone or basso voice, it sounds comb-filtered.) AI music production is about curation, and it can also be about lyric-writing. There is a large gulf between a casual non-musician who accepts one of the first few takes (which is almost all of them) and one who will usually audition at least 50 takes, and sometimes won't stop until they've heard a hundred or more. Each pair of takes (they're generated two at a time) sounds like it's from a different band. You press the button 25 times, that's 50 takes from 25 "bands." It can take days to do this, and by the way, it's NOT good at following instructions. Just learning how to get it to do anything right takes endurance. Then, there's usually editing because a word didn't sound as good as you thought, or it wasn't pronounced right. This can be excruciating. Expect anywhere from 20-50 takes to fix a single phrase in once verse. It can take an hour or more to fix ONE WORD. If you take this seriously, you then download the stems and spend many hours mixing and mastering, because the AI won't do that for you. Its output isn't well-mixed, and it isn't flat either. This takes time to fix. You can also overdub your own instrumentals or vocals, or entirely replace some of the tracks it made, as the output is often indistinct and not enjoyable. You can use Melodyne to convert to MIDI, and some services (Suno) are starting to provide MIDI as well, though it's said not to be very accurate. This is what it means to use AI as a tool in music. It's far, FAR beyond what very nearly everyone who uses AI music services is capable of doing ar all, let alone motivated.