AI can generate sound, but only a person can decide what the work should express — let higher output raise the bar; the listener still decides.