Acoustic Measures According to Speaker Gender Identity: Differences and Correlation With Vocal Satisfaction

Abstract

Objective: To analyze acoustic measures of speech and vowel samples from individuals of different genders and to correlate these acoustic measures with vocal satisfaction. This study aims to provide additional data on acoustic measures, serving as references for clinicians while emphasizing the importance of moving beyond cisgender norms. Additionally, it addresses a gap in the Brazilian context by exploring correlations between acoustic measures and self-perceived vocal satisfaction across diverse gender groups.

Methods: Cross-sectional study. Speech segments and sustained vowel emissions from 47 individuals (11 cisgender women-CW, 11 transgender women-TW, 11 cisgender men-CM, seven transgender men-TM, and seven non-binary individuals-NB) were recorded, and 14 acoustic measures of noise, perturbation, spectral, and cepstral parameters were extracted. Vocal satisfaction was measured using a visual analog scale. Descriptive and inferential analyses were performed using Kruskal-Wallis, Dwass-Steel-Critchlow-Fligner, and Spearman correlation tests, considering P < 0.05.

Results: Significant differences were observed in various acoustic measures, such as the average fundamental frequency (f_o) between CW and CM (P < 0.001), TM (P = 0.015), and TW (P = 0.050); minimum f_o between CW and CM (P = 0.002) and TW (P = 0.050); Jitter between CW and CM (P = 0.013); H1-H2 between CW and CM (P = 0.002); Shimmer between CW and CM (P = 0.046); Acoustic Breathiness Index (ABI) SCORE between CW and CM (P = 0.024); median f_o in speech between CW and CM (P < 0.001), TM (P = 0.009), and TW (P = 0.011); minimum f_o in speech between CW and CM (P = 0.014) and TW (P = 0.031); and maximum f_o in speech between CW and TW (P = 0.050). The measures that correlated with vocal satisfaction were strong positive correlation for spectral emphasis in CW (R = 0.608; P = 0.047) and noise-to-harmonic ratio in TW (R = 0.724; P = 0.012); very strong positive correlation for smoothed cepstral peak prominence in TM (R = 0.847; P = 0.016); and strong negative correlation for coefficient of variation of intensity in NB (R = -0.775; P = 0.041).

Conclusion: Acoustic measures differ according to the speaker's gender in the studied sample. Significant differences were observed in fundamental frequency between CM and cisgender and TW, as well as between cisgender and TW. Differences in fundamental frequency, Jitter, Shimmer, the difference between the first two harmonics, and the were identified across gender groups. The measures correlated with vocal satisfaction differed among the groups. However, these correlations should be interpreted cautiously, as they suggest a relationship but do not imply causation, and their clinical significance requires further investigation.

Keywords: Voice—Voice training—Speech acoustic—Gender identity—Communication.