Towards Speaker Characterization: Identifying and Predicting Dimensions of Person Attribution

A great number of investigations on person characterization rely on the assessment of the Big-Five personality traits, a prevalent and widely accepted model with strong psychological foundation. However, in the context on characterizing unfamiliar individuals from their voices only, it may be hard for assessors to determine the Big-Five traits based on their first impression. In this study, a 28-item semantic differential rating scale has been completed by a total of 33 listeners who were presented with 15 male voice stimuli. A factor analysis on their responses enabled us to identify five perceptual factors of person attribution: (social and physical) attractiveness, confidence, apathy, serenity, and incompetence. A discussion on the relations of these dimensions of speaker attribution to the Big-Five factors is provided and speech features relevant to the automatic prediction of our dimensions are analyzed, together with SVM regression performance. Although more data are needed to validate our findings, we believe that our approach can lead to establish a space of person attributions with dimensions that can easily be detected from utterances in zero-acquaintance scenarios.