Owning Words

 

What AI music lacks is a producer. When you generate music with AI, you don’t have a say in the production, and you just have to accept the sound of it. As I’ve said before, AI music generation is basically a listener activity, like listening to the radio or a playlist, where certain tracks come up–and you like them and you won’t skip over them. But in all the music that I’ve generated, if I were a producer, I would have done them differently. 

What makes a big difference in production is the intimate connection between the lyrics and how the lyrics are sung, as well as the whole vibe of the production. Even one word sung inappropriately will probably ruin the whole piece. I’m sensitive to words, as all singers are. The difference between a singer on an AI track and an actual singer in a studio singing from the lyric sheet is that they can square their own interpretation of the lyrics with their own soul. A singer has to own the words. There are certain vocals that I could never do. I did them in the studio as demos, but reluctantly. I don’t have a powerful voice, so there are certain words that I can’t “own” as other singers can.