Bengali Music Genre Classification Using WaveNet

Music genre classification plays a crucial role in music recommendation and information retrieval. While there have been numerous studies on classifying genres of English music using various machine learning approaches, there is a noticeable absence of similar work for Bangla music classification. Bangla music encompasses a wide range of types and styles that can be classified into different genres. In this research, we focus on six distinct Bangla music genres, including ’Bangla Adhunik,’ ’Bangla Hip-Hop,’ ’Bangla Band Music,’ ’Nazrulgeeti,’ ’Palligeeti,’ and ’Rabindra Sangeet.’ Our dataset consists of 250-300 songs (in .MP3 format) for each genre. By comparing the performances of different models, we demonstrate that our proposed model, using a combination of Principal Component Analysis and K-Fold on the model WaveNet, achieves high accuracy in the multiclass classification of Bangla music genres by comparing with other machine learning and deep learning. This work contributes to the advancement of music genre classification specifically for Bangla music and paves the way for enhanced music recommendation systems and information retrieval in this rich musical tradition.