Making Music

Generative AI is pretty amazing these days. You can practically click a button and have it spit out a full song. Because it looks that easy, people often assume there’s no real creative process behind it. But that’s a pretty naive view.

Sure, you can get a song by hitting a button; however, in my experience, the good ones take hours of iteration.

For me, it starts with the lyrics. I plan the song with the AI, giving it loads of detail and direction: what I want, what I don’t want, the arc, the message. Only then do I build the chorus and wrap the rest of the song around it. Getting the lyrics to feel right can take dozens of revisions, tweaking individual lines until the emotion lands the way I want.

Next is instrumentation and style – figuring out what shirt the song can wear. Not every genre or set of instruments fits. I test different combinations until I find the sound that carries the emotion best.

It’s all deeply iterative. Each revision hones in closer to what the song is meant to be. Even though I’m just a tinkerer, I still want to make something that feels and sounds right. It’s a lot more than just clicking a button.