Penn scientists announce an AI-powered sound engineering audio language model, SmartDj, that turns simply worded commands into complex audio.