AI can generate sound, but only a person can decide what the work should express — judge the result; the listener still decides.