Powerful speech synthesis model.