Auditory modeling of vowels
R. A. W. Bladon, Björn E. F. Lindblom · The Journal of the Acoustical Society of America · 1979
In order to shed light on a specific theory of peripheral auditory representation (drawn from work by Zwicker and recently formalized by Schroeder, Atal, and Hall) we have used data on how listeners match two-formant vowels to four-formant stimuli. We report on two experiments designed to test the hypothesis that such listeners determined their favored F2′ according to a distance measure (adapted from Plomp) formulated in terms of such auditory representations. This theory models the auditory transformation of the acoustic signal as a conversion of the power spectrum into a sones/Bark representation by computing critical-band densities and applying a (basilar membrane-like) frequency smearing function. The computed pseudoauditory vowel distance measure was found to correlate very highly with the listeners judgments of vowel-quality distance obtained from the two experiments described, which involved the comparison of numerous vowel pairs from reported F2′ data. The results suggest that a model of peripheral auditory representation is available which would be useful as a building block in any theory of speech perception.