Music-emotion Recognition Based on Wearable Dry-electrode Electroencephalogram

Despite the success of previous research in emotion recognition using electroencephalogram (EEG), the traditional EEG devices used in those research had limited practicability to be employed in a naturalistic and real-world situation. Accordingly, a new wearable EEG mounted with dry electrodes has been recently developed, yet the feasibility to use it for emotion detection has not been confirmed. In this work, we present a preliminary study of emotion recognition in music listening using the new EEG device. Spectral features were extracted from the selected six electrodes (Fp1, Fp2, LC1, C2, T3, and T4) of the EEG and the support vector machine was employed to classify binary classes of arousal and valence. Our empirical results demonstrate the promise of using the new EEG to recognize emotional states of a human during music listening.