Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

I'm a musician and I've been pretty excited about AI music generation, which seems to stand in contrast to most musicians. So far though, the available AI music gens are just plain terrible, including Suno 4.5. Everything sounds exactly like what you would get if you bought a library of really bad loops and just layered some together. Boring rhythms. Boring harmonic progressions. Boring melody. Boring lyrics. It's just terrible. I just listened to around 100 song on Suno and I didn't find one that was even remotely interesting or even just fun to listen to. I'm not a music snob - I'll listen to everything from Sigur Ros to At the Drive-In to Insomnium to Ella Fitzgerald to Taylor Swift to Dua Lipa. I just want something that can grab my attention, show me something interesting, and generate a feeling.

Here's the thing about music - what makes a good song is a sense of unexpected familiarity. In other words, it's based on a set of broad, familiar rules (music theory), but surprises you by implementing those rules in a way that's extremely unique and novel. Because generative AI is kind of like an averaging machine, rounding off the outliers, what you get is an example of the rules in the most boring, typical way possible. It completely lacks the element of surprise, and I'm not sure it will ever be capable of that essential element of surprise because of how it fundamentally works. If you want to generate a bunch of background elevator music, however, this seems to be extremely useful for that.




Consider applying for YC's Fall 2025 batch! Applications are open till Aug 4

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: