A deep music genres classification model based on CNN with Squeeze & Excitation Block

With the development of mobile terminals and Internet technology, people have increasingly convenient mediums to obtain digital music. However, complex music genres and massive music libraries have brought great challenges to music information retrieval. Music genres are high-level labels for music information, which would consume a lot of time and resources when manually tagged. This paper proposes a new model: in order to fully mine the latent information hidden in the input spectrum graph, we build a music genre classification system based on the convolutional neural network that includes Squeeze & Excitation Block (SE-Block), and then use Bayesian optimization to search the best parameters of SE-Block. Finally, we choose the GTZAN dataset for experiments and achieved a classification accuracy of 92%, which is significantly better than most previous research work.