Masculinized or Feminized? Discriminant Analysis of Postmenopausal Women’s Voices

Abstract

This study investigates the degree of vocal variation between men, pre- and postmenopausal women. The sample comprised 108 volunteers aged 18–66 y., divided into control and validation group. Each participant was subjected to voice recordings of five sustained vowels. Acoustic parameters were extracted using Praat software. The most significant parameters in intergroup correlation between canonical discriminant function and acoustic variables were: fundamental frequency (f0), shimmer, harmonics-to-noise ratio (HNR) and intensity. Premenopausal female voices were labeled with 97% correctness and male voices with 95.5% correctness. Interestingly, 65.5% of postmenopausal women were accurately classified as female voices and on average they had lower vocal pitches compared to premenopausal women. The differences in male and female voices are probably due to the difference in the size of the larynx and the length of the vocal cords. Hormonal changes during menopause may affect, but not significantly, the morphology of the laryngeal structures which develop during childhood and adolescence.