A transformer-based audio generation model that creates highly realistic speech, music, and sound effects.