Music-Evoked Emotion Classification from EEG: An Image-Based CNN Approach

Music has a profound impact on human emotions, and the ability to automatically recognize the emotional content of music has numerous applications in fields such as entertainment, healthcare, and human–computer interaction. In this study, we propose a novel approach for Music-Emotion Recognition based on Electroencephalography (EEG) signals, leveraging the power of Convolutional Neural Networks (CNNs). EEG data were collected from participants while they listened to two sets of Indian and Western music clips, pre-rated for a range of perceived emotions. The EEG signals were pre-processed and converted into images. These images were used as input to the CNN architecture, which was specifically designed to capture relevant patterns and features related to music-induced emotions. To train and evaluate the CNN model, we employed a cross-validation framework, achieving high accuracy in emotion recognition across multiple emotional categories. Results revealed that the highest emotion prediction accuracy was achieved using VGG16 architecture, especially during the middle part of the music clips. Overall emotion classification accuracy for Western clips was higher than Indian clips, and exciting emotion featured a higher consistency in neural responses throughout the whole clip duration compared to other emotions.