Analysis of speech features and personality traits

Abstract

Voice signal has been widely investigated to characterize mood and emotional states. A further interesting dimension could regard the personality traits. The relationship between personality traits and specific speech features is known, however this topic requires further investigation. Specifically, most studies are focused on perceived personality traits, without adopting dedicated personality tests. Moreover, the relationship among speaker personality traits and specific speech features have still to be clarified. In this study, a correlational analysis between some speech-related features and the personality traits, as described by the Zuckerman-Kuhlman model and the Toronto Alexithymia Scale, is performed. An experimental protocol, consisting of two structured speech tasks, was administered to eighteen healthy subjects. Speech features were estimated to describe fundamental frequency (F0) and voice quality related features from whole speech recording and tilt-related features, describing F0 dynamics at voiced segment level. Significant correlations among personality traits and speech features were observed using both feature sets. Interestingly, the adopted speech task was found to influence the obtained results. Specifically, no feature reports the same significant correlation in both adopted tasks. The impact of personality traits and speech production studies on the characterization of mental disorders and the estimation of emotional/mood state of the speaker are discussed.