An acoustic-perceptual method for the quantitative evaluation of hypernasality

A computer implementation is proposed as a method for assessing the quality of an individual's speech from a perceptual and communicative point of view. The input to this procedure is a standardized set of speech samples (vowels and selected consonants) designed to permit an evaluation of the speaker's phonology. The output is a numerical specification of the functional adequacy of the target values of the phonetic segments under analysis. The computer program consists of three parts: a simulated auditory analyzer, the output of which is a spectral representation of each test item; an algorithm that computes a psychoacoustically based measure of distinctiveness or perceptual distance for each pair of test spectra, and a routine for comparing, in terms of this measure, the test items with the corresponding reference items drawn from a population of normal speakers or a recording made before patient intervention.