Ensemble Classifier based on Decision-Fusion of Multiple Models for Speech Emotion Recognition

This paper proposes an ensemble classifier based on decision-fusion of multiple SER (Speech Emotion Recognition) models. The one of the multiple SER models used in this work is a typical categorical learning model for classifying the emotion labels, while the others are A/V (Arousal/Valence) models that recognize multiple A/V states based on the Russell's A/V emotion space. The evaluation performed in this work shows that the SER accuracy of the proposed ensemble classifier that combines each output of categorical model and A/V models is improved compare to the result when each SER model is applied separately.