The psychoacoustics of noise vocoded speech: a physiological means to a perceptual end.

Noise vocoded speech tokens produce temporal patterns in the ensemble response of the auditory nerve similar to those of their naturally produced counterparts [Loebach, J.L., Wickesberg, R.E., 2006. The representation of noise vocoded speech in the auditory nerve of the chinchilla: Physiological correlates for the perception of spectrally reduced speech. Hear. Res. 213 (1-2), 130-144]. Moreover, the degree of pattern similarity increased as more noise bands were used to synthesize the vocoded stimuli, suggesting a relationship between the patterns that these stimuli evoke in the auditory nerve and their recognition by human subjects. In order to make a direct comparison between the psychoacoustic and physiological domains, the present study obtained the perceptual identification scores for these stimuli. A set of 120 stimuli containing the 16 tokens of interest was presented to 30 young normal hearing subjects, who identified the tokens in a closed set task. Overall, the perceptual identification of the tokens increased in accuracy with the addition of noise bands. The neural pattern similarity was quantified using dynamic time warping, and correlated with the perceptual identification scores for the target stimuli of interest. A significant linear relationship between the pattern similarity and perceptual identification scores was found, such that as neural pattern similarity increased, the accuracy of stimulus identification also increased. These findings suggest a possible physiological substrate for the recognition of noise vocoded consonants.