Carlo Giovannella, Daniele Floris, Andrea Paoloni
pp. 102-111, download
(https://doi.org/10.55612/s-5002-015-008)
Abstract
This article presents the first attempt to investigate the existence of possible quantitative correlations among the physical characteristics of emotional portrayals and the emotions perceived by humans during their listening. Our aim was: a) to design and develop a new investigation procedure/protocol; b) to obtain information useful to recognition and synthesis of emotions conveyed by the human voice. Our results, obtained on a subset of the emotional portrayals contained in the corpus EMOVO, show that, apart from well known qualitative indications, it is also possible to observe clear quantitative trends for some couples of ‘emotionsignal characteristics’ as function of the recognition rate of the emotion.
keywords: voice emotional portrayals, perception of emotion, emotion recognition and sinthesis, EMOVO