Studying vocal social attractiveness by re-synthesis – results from two student projects applying acoustic morphing with Tandem- Straight

Authors: Benjamin Weiss, Anabell Hacker, Cleopatra Moshona, Frederic Rudawski, Matthias Ruhland

Abstract:

Acoustic analysis and synthesis experiments provide meaningful methods to study vocal processes of social cognition. Common attributions and evaluations of person’s speech include personality, emotions, and social attitudes. Recent advances in speech coding allow improving the synthesis methodology by defining valid parameter target values without much manual analysis effort. In two student projects, the de-factor standard for speech coding/morphing, Tandem-Straight, was applied to conduct experiments on vocally manifested social cognition processes. Hypothesisbased stimuli were produced on four, respectively five steps in a bi-directional fashion. The evaluations took about 30 min. for each participant. Each group conducted a qualitative analysis for stimulus selection and a quantitative pre-test to obtain social evaluations for hypothesis formulation. During this preparation, both groups decided to select female speakers, as only these provided sufficient variations in acoustics and social ratings. Results show an impact of a darker spectrum on liking ratings for 47 participants, as well as an effect of different, eventually lower, F0 on dominance attribution for 51 participants.