Music Emotion Classification using Harris Hawk Optimization based LightGBM Classifier

This research investigates the relationship between music and human emotions, focusing on emotion classification in Turkish music and the role of acoustics in music emotion recognition (MER). Leveraging a unique dataset stored in the UCI-MLR, the study employs machine learning techniques, including Linear Discriminant Analysis and Gradient Boost Classifier, to build a robust model. The accuracy enhancement is primarily driven by the feature selection process, which is powered by the Harris Hawks optimizer. This optimizer, when used with the LightGBM model, enables a more precise selection of features, thereby significantly improving the model's accuracy. The research achieves notable accuracies—81.25% for Linear Discriminant Analysis and 83.75% for the Gradient Boosting Classifier—surpassing previous records. The paper discusses findings in the context of affective computing, emphasizing potential applications in music recommendation systems and therapy, concludes with a comprehensive analysis, and outlines future possibilities. This research contributes to advancing MER, laying the groundwork for innovations in diverse domains, including the potential integration with IoT-based intelligent systems for real-time emotion analysis and personalized music experiences based on camera frames.