I have 60000 generated audio samples of which the MFCC features where calculated (13feautres x 44 steps). I want to use a neural network to learn a mapping between the MFCC
