Classification of emotions by Artificial Neural Networks: a comparative study

EMANUEL TELLO; PEREZ ELISA; FERNANDO MUÑOZ; FLAVIO ROBERTI; DANIEL PATIÑO; NATALIA M. LÓPEZ

Medellín

Congreso; XVII Latin American Conference of Automatic Control (CLCA 2016); 2016

Universidad EAFIT

Every emotion evidences a biological sign while predisposes the body to a different kind of response. In Human-Computer interaction area, the voice recognition techniques are widely used in text engines. In this context, the automatic emotion recognition of the speech aims at identifying the emotional or physical condition of a human being from his voice. Both emotional and physical states of a speaker are included in so-called paralinguistic aspects. In this work we used a speech database containing 7 different emotions in which 4 emotions were selected and 12 features were extracted. Emotions were classified by different types of neural networks in order to compare the efficiency of them to discriminate different moods.