Music Genre Classification

In this notebook, we attempt to classify music signals for which we have two representations ; MFCCs (https://en.wikipedia.org/wiki/Mel-frequency_cepstrum) and mel-spectograms using neural networks in Pytorch.We start with a FNN and then proceed to a CNN experimenting on optimizers,activation functions, learning rate schedulers,pooling and padding,dropout,early stopping and batch normalization.Our best model presents 79.8% accuracy on test set.