A machine can predict patterns, but taste begins where prediction ends — let higher output raise the bar; then the music can speak.