WGAN neural network generated audio.

I applied the WGAN-GP algorithm to raw audio data using 1D convolutions. After 16 hours of training it came up with this. Still needs a lot of work but at least it got somewhere.