Contribution of voice fundamental frequency and formants to the identification of speaker's gender

Abstract

Identification of gender from speech sounds has been found to rely on speakers’ voice fundamental frequency (F0) and formant frequencies. The present study aims at examining the contribution of F0 and formants to the correct detection of speaker’s gender. Based on the vowel sustained by a male and female speaker, 200 vowels were synthesized with a range of F0-formant combinations. The synthesized vowels were presented to 28 native Cantonese-speaking listeners to judge the perceived speakers’ gender for each of the synthesized stimuli. Results revealed that F0 was the primary cue for speakers’ gender perception while formants contributed little. The cutoff F0 values for male and female identification were found to be 162.01 Hz and 204.97 Hz, respectively. When F0 was below 162.01 Hz or above 204.97 Hz, listeners reliably and correctly identified the speakers as male or female, respectively.