Identifying Emotions in Speech Patterns: Adopted Approach and Obtained Results

Abstract

There are an increasing number of studies in the area of Human-Computer Interaction (HCI) which bear witness to the importance of taking note of the emotional factors involved when interacting with computing systems. Once the emotions of the users are known, it is possible for artificial agents to influence these feelings with a view to interacting with them and stimulating their routine or specific activities on a day-to-day basis. However, providing computing systems with the capacity to recognize and interpret the emotions of their users, is one of the great challenges in the area of HCI. For this reason, this article adopts an approach based on the Classification Committee which is concerned with identifying and classifying emotions on the basis of the motor expression of the users (in this case the voice). When this procedure is compared with the classical approaches adopted in the literature, the results show that, as well as achieving a high degree of accuracy, the proposed model maintains a good level of consistency when identifying the emotions of the users.