Acoustic Variables as Predictors of Vocal Attractiveness

Abstract

The primary goal of this paper is to provide a full-fledged account of vocal attractiveness from phonetic perspectives. Despite its importance in phonetic and social domains, previous research on vocal attractiveness remains far from comprehensive. Therefore, due to the absence of comprehensive research on vocal attractiveness, this paper seeks to address the following research questions: a) what is the most reliable acoustic variable for predicting vocal attractiveness?, b) does sexual dimorphism have an impact on vocal attractiveness?, and c) how does vocal attractiveness judgment differ according to gender? Production and perception experiments were conducted to answer these questions. In the production experiment, three sentences from sixty Korean native speakers (37 males and 23 females) were recorded. The sentences included the three point vowels a, u, and i in the form of carrier phrases. In the perception experiment, twenty native Korean listeners (10 males and 10 females) scored the anonymous voices recorded during the production experiment using a seven-point scale with endpoints labeled strongly disagree and strongly agree in response to the statement: Do you find the following voice attractive? Acoustic measures were analyzed using Praat and VoiceSauce in MATLAB. The full set of measures computed in this research is as follows: duration, H1-H2, Cepstral Peak Prominence (CPP), energy, energy range, Harmonics-to-Noise Ratio (HNR), vowel space, formant dispersion, f0, f0 maximum, f0 minimum, and pitch range. A multiple linear regression with a stepwise method and independent t-test were carried out for the statistical analysis. Overall results were as follows: as an answer for research question #a, duration was the most reliable acoustic measure for predicting vocal attractiveness in males and formant dispersion was the most reliable acoustic measure for predicting vocal attractiveness in females. Males with a slower speech rate and a low f0 were regarded as having an attractive voice. Females with a) feminine voices, b) clear and lucid voices, and c) voices with a higher maximum value of f0, lower minimum value of f0, and reduced pitch range were rated as more vocally attractive. The results for research question #b revealed that sexual dimorphism has a distinct effect on vocal attractiveness perception. The more masculine a mans voice (the lower their f0), or the more feminine a womans voice(the higher their formant dispersion), the higher their level of vocal attractiveness was perceived to be. In response to research question #c, the research showed that male raters tended to give opposite-sex benefits in assessing vocal attractiveness while female raters did not, instead being observed to have given higher scores to the same-sex.