Music Latent Representation and Disentanglement for Genre Classification

Published:

In this project, we used unsupervised representation learning techniques to compress musical samples into a disentangled low-dimensional representation to classify music genre. The training and evaluation of the models were done using the FMA public dataset.