Perceptual similarity judgments of voices: Effects of talker and listener language, vocal source acoustics, and time-reversal

Abstract

Listeners demonstrate reliable perceptual biases favoring voices speaking their native language versus those speaking a foreign language. This “language-familiarity effect” for voice processing has been found even for perceptual similarity judgments: Voices speaking listeners’ native language sound less alike, even when those recordings have been rendered incomprehensible via time-reversal. Here, we sought to replicate and extend this finding of linguistic effects on voice similarity. Native English- and Mandarin-speaking listeners (both N = 40) rated the perceptual similarity of voices speaking English or Mandarin (both N = 20) for either time-reversed or forward speech. Both listener groups tended to find Mandarin-speaking voices more dissimilar, but this effect was reduced for English-speaking listeners, especially for forward speech. Perceptual similarity judgments of voices were always highly correlated between listener groups and between forward/time-reversed speech. Acoustic measurements (voice fundamental frequency mean and variance, local jitter, harmonics-to-noise ratio, speech rate, and formant dispersion) were also made to ascertain how listeners’ perceptual similarity judgments were related to speech acoustics and whether these relationships differed under time-reversal or across listeners’ or talkers’ native language. Overall, these data suggest that, while native language may influence perceptual dissimilarity of voices, the magnitude of these effects tend to be very small.