The combined effects of distortions on vowel identification.

Abstract

The effects of intrinsic and extrinsic distortions on vowel perception were evaluated. Potential sources of distortion were high presentation level, background noise, stimulus ambiguity, and hearing status. Vowel identifications were collected for three sets of synthesized vowels that varied in both F1 and F2 (front and back vowels), or F2 and F3 (central vowels). The systematic variation in formant frequencies produced stimulus sets that contained both clear, unambiguous category tokens as well as relatively ambiguous tokens that could be identified as more than one vowel category. As overall presentation level increased from 75–90 dB SPL, identification by normally hearing listeners became less consistent for ambiguous stimuli. Additional reductions in categorization precision occurred in the presence of background noise. The effects were highly dependent on stimulus set, with the largest effects observed for the central vowels. Data from hearing-impaired listeners at high presentation levels and in noise will also be reported. Evaluation of response patterns in the presence of combinations of these sources of distortion will help explore the difficulty people with hearing loss have processing speech in adverse listening conditions. [Work supported by OHSU Medical Research Foundation, Portland, OR.]