Linguistic and personal influences on speaker variability

Abstract

Our previous studies examined the manner in which within- and between-speaker acoustic variability in voice follow patterns determined by biological factors, the language spoken, and individual idiosyncrasies. To date, we have analyzed data from speakers of English, Seoul Korean, and Hmong, which differ in whether they contrast phonation type and/or tone. We found several factors that consistently account for acoustic variability across languages, but also factors that vary with phonology. The present study adds Gujarati (which contrasts breathy with modal phonation) and Thai (a tone language without contrastive phonation) to this work. We hypothesize that F0 will emerge from analyses of Thai, as it did for Hmong and Korean (but not for English), and that differences in the amplitudes of lower harmonics will emerge for Gujarati, as occurred for Hmong, but not for English or Korean. We further hypothesize that two factors—the balance of high-frequency harmonic and inharmonic energy and formant dispersion—will emerge as the most important factors explaining acoustic variance for these new sets of speakers and languages, as they have in our previous studies. Such a result would be consistent with the view that speaker variability is governed by both biological and linguistic factors.