Construction of a database of emotional speech using emotion sounds from movies and dramas

Abstract

In this study, an emotional speech database called Hanbat Emotional Database (HEMO) was constructed using movie and drama scenes in which emotion is abundantly expressed by professional actors. HEMO consists of 454 speech samples classified into seven emotion categories such as anger, happiness, sadness, disgust, surprise, fear, and neutral. In order to evaluate the performance of HEMO, consistent experiments were conducted based on HMM (Hidden Markov Model) and GMM (Gaussian Mixture Model) for both HEMO and the Berlin Emotional Speech Database (EMO). HEMO showed better results than EMO with a positive recognition rate of 78.89%.