Research on Architecture for Long-tailed Genre Computer Intelligent Classification with Music Information Retrieval and Deep Learning

Abstract

In this paper, we propose a Musical Attention Network (MAN) architecture for long-tailed, imbalanced music genre classification which is often ignored and quite prevalent in Music Information Retrieval (MIR). Here, the challenge is to classify the genre of long-tailed music accurately. Inspired by the recent progress in NLP, the proposed model can take advantage of genre correlations to better identify informative segments. Comprehensive experimental results demonstrate our model brings significant improvement comparing with other music genre classification models on a large-scaled benchmark dataset.