Time to Frequency Domain Mapping of the Voice Source: The Influence of Open Quotient and Glottal Skew on the Low End of the Source Spectrum

Abstract

This paper explores the mapping of time and frequency domain aspects of the voice source, focussing on the low end of the source spectrum. It refines and extends an earlier study, where the LF model was used to explore the correspondences between the open quotient (Oq), glottal skew (Rk) and harmonic levels of the source spectrum, including the H1-H2 measure, widely assumed to reflect differences in Oq. Here we use a different model (the F-model) as it better reflects the effective open quotient and glottal skew in certain conditions. As in the earlier study, a series of glottal pulses were generated, keeping peak glottal flow constant, while systematically varying Oq and Rk. Results suggest that the effects of Rk on the low harmonics is considerably less than estimated in the earlier study, and its main impact is on the level of H2 (and consequently H1-H2) when Oq is relatively high. The conclusion remains that the H1-H2 is not simply a direct reflection of Oq. However, for Oq values of up to about 0.6, it maps closely to H1-H2: beyond this point, H1-H2 reflects a more complex interaction of open quotient and glottal skew.

Voiced aspirates with mixed voicing in Yemba, a Grassfields Bantu language of Cameroon

“…The particular set of measures we adopt here is commonly used in previous literature on voice quality, but we omit the very commonly used measure H1 * -H2 * , opting to use H1 * -A3 * in its place as a measure of spectral tilt. This decision was made in light of recent research showing that H1 * -H2 * does not directly reflect laryngeal articulation(Chai & Garellek 2022;Gobl & Ní Chasaide 2019).https://doi.org/10.1017/S002510032300018X Published online by Cambridge University Press…”