Speaker Normalization in Speech Perception
Speech produced by different people varies acoustically because of individual differences in vocal tract physiology. This acoustic variation in the “same” words of language presents a practical problem for linguists and engineers, as well as a theoretical problem in the study of speech perception. This chapter reviews some practical vowel-normalization methods and discusses the perceptual processes that listeners may use to accomplish speech recognition in the face of talker variation, with a general overview of cortical organization for talker normalization and speech perception. The chapter reviews evidence that there are aspects of auditory processing that help remove talker differences during speech perception - in auditory spectral encoding; and in contrast coding in an auditory frame of reference that seems to apply before lexical processing begins.