Genre Classification from Bangla Music Lyrics: Dataset and Classifiers

Music is one of the most important means of expressing people’s emotions and love. Music can be divided into different genres. Interest in music varies from person to person. Different people are interested in different genres of music. So categorizing music into different genres is very important to enhance the user experience of music streaming platforms. In this paper, we have presented a Bangla music dataset consisting of more than 10500 song lyrics distributed across 11 genres. We have used classical machine learning algorithms (Support Vector Machine, Random Forest), deep learning-based models (BanglaBERT, mBERT, Bloom2, etc.), and a CNN model for genre classification based on song lyrics. Among all these models, BanglaBERT outperforms all of them. We achieved the highest accuracy of 92.00% and an F1 score of 92.00%.