“The concept of a unified model capable of generating speech, music, and sound effects.”