The main purpose of this work is to try to apply image style transfer methods on audio signals.

Explanations for features detected by CNNs are still missing. We introduced an external Knowledge base to provide an explanation for CNN features to give a glimpse inside the CNN black boxes.