AN EMOTIONAL FEATURE ANALYSIS IN SPEECH SIGNALS
Mie Sato, Miyuki AKATSUKA, Yoshito Mekada, Masao Kasuga · KANSEI Engineering International · 2002
To achieve a comfortable speech interaction with a computer, it is expected that the computer can speak to us with human emotions and can recognize our feeling from our speech. In this paper, we ascertain physical features included in our conversation. We focus on the most three major emotions of the basic emotion and collect speech samples of each emotion. The speech signals with these emotions are compared with those which have no emotion by perceptual experiments, and are analyzed for their physical features. Furthermore, we perform a discriminant analysis with four discriminative predictors which indicate quantitative features of the emotions in the speech signals. Our analysis effectively shows features of the speech signals with the emotions: ‘surprise’, ‘sorrow’ and ‘joy’.