Abstract
Both voice gender perception and speech perception rely on neuronal populations located in the peri-sylvian areas. However, whilst functional imaging studies suggest a left vs. right hemisphere and anterior vs. posterior dissociation between voice and speech categorization, psycholinguistic studies on talker variability suggest that these two processes share common mechanisms. In this study, we investigated the categorical perception of voice gender (male vs. female) and phonemes (/pa/ vs. /ta/) using the same stimulus continua generated by morphing. This allowed the investigation of behavioral differences while controlling acoustic characteristics, since the same stimuli were used in both tasks. Despite a higher acoustic dissimilarity between items during the phoneme categorization task (a male and female voice producing the same phonemes) than the gender task (the same person producing 2 phonemes), results showed that speech information is being processed much faster than voice information. In addition, f0 or timbre equalization did not affect RT, which disagrees with the classical psycholinguistic models in which voice information is stripped away or normalized to access phonetic content. Also, despite similar average response (percentages) and perceptual (d') curves, a reverse correlation analysis on acoustic features revealed that only the vowel formant frequencies distinguish stimuli in the gender task, whilst, as expected, the formant frequencies of the consonant distinguished stimuli in the phoneme task. The 2nd set of results thus also disagrees with models postulating that the same acoustic information is used for voice and speech. Altogether these results suggest that voice gender categorization and phoneme categorization are dissociated at an early stage on the basis of different enhanced acoustic features that are diagnostic to the task at hand.
Highlights
The human voice is what we use to communicate on a daily basis and there is evidence that voices are “special” in that they are processed differently from other stimuli in the brain (Belin et al, 2000, 2004; Petkov et al, 2008)
Despite fewer studies looking at voice perception neuroanatomy than speech perception neuro-anatomy, two main brain regions have been identified as supporting voice and voice gender perception
On one hand, faster Reaction time (RT) observed in the phoneme task than in the gender task, along with the absence of effect of f0 or timbre equalization, suggest that voice is not stripped away from speech to access phonemic content
Summary
The human voice is what we use to communicate on a daily basis and there is evidence that voices are “special” in that they are processed differently from other stimuli in the brain (Belin et al, 2000, 2004; Petkov et al, 2008). Many factors are likely to be involved in the perception of vocal sounds, both linguistic and paralinguistic. One such factor is the pitch, with the pitch height (the perception of the fundamental frequency f0 of the vocal fold vibration) being its’ main characteristic. Timbre (influenced in particular by the distribution of energy across frequencies, as observed in the power spectrum of the sound) is another factor involved in the perception of vocal sounds, and is perceived as the characteristic quality of a sound. The categorization of voice gender appears to depend on the right voice selective areas to encode acoustical dissimilarity (Charest et al, 2012). The frontal cortex and in particular the bilateral inferior frontal regions, seem to be important in the encoding of perceived ambiguity and to carry out categorical perception (Charest et al, 2012)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
