Computational Estimates of Pitch Height and Pitch Strength in Dysphonic Voices

Abstract

<h3>Research Objectives</h3> To evaluate the utility of automated computational estimates of pitch height and pitch strength in characterizing dysphonic voices that have glottal (i.e., true vocal folds) and supraglottal (e.g., false vocal folds), vibratory sources (GVS and SGVS). Furthermore, the utility of such pitch-based measures (Zwicker & Fastl, 1990) in characterizing and differentiating these voice signals that vary in periodicity (Type I, II, and III) was also tested (Titze, 1994). <h3>Design</h3> Repeated measures. <h3>Setting</h3> University. <h3>Participants</h3> Dysphonic voice samples (sustained /a/ and sentence) from 30 adults with GVS (10 per signal type) and 30 adults with SGVS (10 per signal type) were selected from a large database through stratified-random sampling. Pitch height and pitch strength were estimated from Auditory-Sawtooth Waveform Inspired Pitch Estimator Prime (Aud-SWIPE', Camacho, 2012) algorithm and compared with conventional fundamental frequency (f0) estimates from PRAAT software (Boersma, 2022). <h3>Interventions</h3> Not Applicable. <h3>Main Outcome Measures</h3> It is hypothesized that dysphonic voices with SGVS will have lower pitch height and pitch strength than GVS due to the inherent differences in anatomical structures of vibration and their associated physiological properties (Alipour & Karnell, 2014; Schwarz et al., 2011). It is also hypothesized that there will be a significant difference in pitch height and pitch strength between the Type I, II, and III voice signals varying in periodicity (Anand et al., 2021). <h3>Results</h3> Pitch height and pitch strength were higher in GVS compared to SGVS. Estimates of pitch height and f0 from were highly correlated and were in close agreement for Type I signals compared to other signal types where PRAAT was unable to provide f0 trace for some voices. Pitch strength was highest for Type I signals and lowest for Type III signals. <h3>Conclusions</h3> Pitch height and pitch strength can be measured in all dysphonic voices irrespective of signal periodicity. <h3>Author(s) Disclosures</h3> Supraja Anand receives salary from the University of South Florida.