Music Genre Classification with Transformer Classifier
Yingying Zhuang, Yuezhang Chen, Jie Zheng
Abstract
Music Genre Classification is a significant and practical field of Music Information Retrieval. Deep learning is increasingly being applied to Music Genre Classification for two main reasons. Firstly, it avoids the manual selection of audio signal features. Secondly, the hierarchical topology is consistent with the layering structure of music in the time and frequency domains. The existing work used CNNs or RNNs or their combination to classify music genre, which have limitations on learning dependencies between distant positions in a sequence. Inspired by an advance in Natural Language Processing (NLP), we designed a Transformer classifier. The Transformer classifier analyzes the relationship between different audio frames well and achieves better performance in Music Genre Classification.