Keywords---deep learning, audio, music, generation, expressive timing
and dynamics