EEG-Based Emotion Detection and AI-Generated Music: A Computational Approach to Personalized Emotion Modulation

This research presents a novel EEG-based emotion detection framework using Convolutional Neural Networks (CNNs) and AI-generated music to regulate emotional states. EEG signals are preprocessed using ICA and high-pass filtering, followed by Power Spectral Density (PSD) feature extraction. The proposed CNN classifier achieved a classification accuracy of 90.2%, outperforming traditional models such as SVM and Random Forest. Based on detected emotional states, the system generated personalized music, which led to a measurable increase in Alpha band power (by 37.8%) and a decrease in Beta/Gamma bands (by 13.1% and 10.7%), indicating a shift toward relaxation. Subjective feedback also confirmed 85.2% stress reduction and 78.6% improvement in happiness. This framework highlights the potential of EEG-informed AI music in emotional regulation.