Abstract:

Common experience suggests that, to a certain extent, people can be recognized by their speech, even over the limited bandwidth of a telephone line. Such talker individuality is at least partly determined by anatomical features such as vocal-fold and vocal-tract geometry. Additionally, linguistic factors such as dialectal and speech community membership give rise to noticeable variation across talkers. In this paper two acoustic/phonetic studies which sought to hold these sources of variation constant will be presented. The first is a study of the relative strengths of (perceptually trading) acoustic cues for the /s/--/sh/ distinction in English. The talkers in this study were a dialectally homogeneous group of female college undergraduates. Substantial individual differences in the relative strengths (in a discriminant analysis) of acoustic cues such as F2 transition onset, frequency of the first fricative pole, and duration of the fricative noise were found. The second study explores the patterns of pronunciation exhibited by identical twins. Previous research suggests that identical twins have virtually identical long term average spectra, and that their voices are difficult to discriminate. The present study is designed to look in closer detail at twins' patterns of pronunciation in diphthongs, voice onset time, and cue trading.