Perceptual consequences of spectral manipulations for cisgender and transgender speakers

Abstract

Fundamental frequency and vowel formant frequencies are reportedly the most salient acoustic-phonetic cues used to judge speaker gender. However, the influence of these cues on listeners' extraction of gender and lexical information from spontaneous speech is poorly understood. This study evaluated the perceptual impacts of modifying fundamental frequency and formant frequencies in spontaneously produced speech by cisgender and transgender individuals. Spontaneous speech samples from 12 transgender (6 transfeminine and 6 transmasculine) and 12 cisgender (6 male and 6 female) speakers were included in three tasks: a speech-in-noise intelligibility task, a two-alternative forced-choice gender identification task, and a masculinity/femininity rating task, in one of two conditions: an unmodified condition or a modified condition in which fundamental and formant frequencies were manipulated to fall within either prototypical cisgender male/female range for the transgender speakers or an intermediate, gender-ambiguous range for the cisgender speakers. Results suggest shifting fundamental frequency and formant frequencies toward prototypically male/female ranges in connected speech can effectively alter listeners' gender identification judgements, but these variables alone may not be sufficient to alter perceived vocal masculinity/femininity to within typical cisgender ranges for transgender speakers. Further, strategies used by transfeminine speakers to achieve gender-congruent voice may negatively affect speech intelligibility.