Characterizing the Effects of Disorder Type, Vowel, and Pitch on the Temporal Variability of Electroglottographic Signals in Dysphonic Voices.

OBJECTIVES

Electroglottographic (EGG) signals are inherently complex in the time domain, yet conventional measurement techniques frequently fail to capture their temporal variability. This study introduces curvEGG, a novel approach designed to quantify the time-dependent complexity of EGG signals based on instantaneous curvedness. As a proof of concept, curvEGG is applied to characterize glottal source complexity across variations in vowel, pitch, and voice disorder type.

STUDY

DESIGN

Parameter development and exploratory retrospective case-control study.

METHODS

CurvEGG was derived by computing the integrated squared second derivative of a preconditioned EGG segment, representing its total time-dependent variation, and normalizing this value against that of an ideal EGG archetype with a matched fundamental frequency. EGG recordings from the Saarbrücken Voice Database were analyzed for 30 women (mean age = 33 years, SD = 9.3), including ten vocally healthy, ten with hyperfunctional dysphonia, and ten with secondary hyperfunctional dysphonia accompanied by vocal fold pathology. Sustained vowels [i], [u], and [a] produced at normal-, low-, and high-pitch conditions were examined. Linear mixed models assessed effects of pathology and vowel in each pitch condition; associations with extremely high curvEGG values were evaluated using Fisher's exact tests of contingency tables.

RESULTS

Low-pitched phonation most effectively characterized healthy and dysphonic voices, with curvEGG increasing progressively from the healthy to the secondary hyperfunctional dysphonia group. Normal- and high-pitch conditions revealed complex interactions between vowel type and pathology, with generally higher EGG complexity for [a] compared with [i] and [u]. Extremely high curvEGG values were typically associated with pathological voices at low-pitch (P < 0.0001).

CONCLUSION

Based on the present findings using curvEGG, it was demonstrated that temporal variations in the EGG offer encouraging signs in characterizing dysphonia in the time domain, particularly under low-pitch conditions. A larger clinical study of voice-disordered adults is needed to confirm the diagnostic potential of curvEGG and its further ameliorations.