Toxicity Classification on Music Lyrics Using Machine Learning Algorithms

Abstract

Music lyrics have a broad scope of impacts on our day-to-day life. The connection between music and the cerebrum has been extensively studied as far as feeling and intellectual interaction. From school children to strict adherents, the audience has the right to taste great music. For example, men presented with physically rough verses tend to more generalized perspectives toward ladies. Listening to particularly toxic or nontoxic songs can affect our mood. Music recommendation system follows different features based on the user’s historical data. The listener’s mode could be improved if the recommendation system filters out toxicity. In this study, we classify lyrics in terms of toxicity and nontoxicity from different music genres and artists using machine learning (ML) algorithms. The toxicity and nontoxicity have been measured using high valence and low valence. From the results, we found that Random Forest (RF) is a much more effective toxicity classification classifier, giving an overall accuracy of 93%.