The role of speaker gender identification in relative fundamental frequency height estimation from multispeaker, brief speech segments.

A perception experiment was conducted to evaluate the proposal that speaker gender identification underlies the ability to estimate relative F0 height from multispeaker speech without cues typically present for speaker normalization. The Mandarin syllable sa was processed to generate fricative, vowel, and fricative-vowel stimuli. Both Mandarin and English listeners identified gender above chance from vowel and fricative-vowel stimuli. Fricative-vowel stimuli were identified more accurately than vowel stimuli, which were identified more accurately than fricative stimuli. Accuracy was comparable between Mandarin and English listeners, but reaction time showed distinct patterns. The perceptual evidence supports the idea that gender identification contributes to relative F0 height estimation.