Effect of vowel context in cepstral and entropy analysis of pathological voices

Abstract

This is the accepted version of a paper published in Biomedical Signal Processing and Control. This paper has been peer-reviewed but does not include the final publisher proofcorrections or journal pagination.

Effects of Fundamental Frequency, Vocal Intensity, Sample Duration, and Vowel Context in Cepstral and Spectral Measures of Dysphonic Voices

“…In tests of accuracy, both CPP and CPPS derived from SV and vowels from connected speech showed a similar discriminatory power for the detection of perceptual dysphonia (Heman-Ackah et al, 2003;Watts & Awan, 2011). However, Selamtzis et al (2019) found a better discriminatory power in extracted vowels from speech.…”

“…CPP and CPPS have been described as more reliable in dysphonic voices, since they do not rely on correct recognition of fundamental frequency (F0) and sound pressure level (SPL) to the same extent as so-called traditional acoustic measures such as jitter and shimmer (Halberstam, 2004;Heman-Ackah et al, 2002;Patel et al, 2018). Moreover, CPP and CPPS showed high correlation and accuracy indices for detection of perceptual dysphonia in sustained vowels and also in connected speech (Awan, Roy, Jette, Meltzner, & Hillman, 2010;Brinca, Batista, Tavares, Goncalves, & Moreno, 2014;Delgado, León, Jiménez, & Izquierdo, 2017;Delgado-Hernandez, Leon-Gomez, Izquierdo-Arteaga, & Llanos-Fumero, 2018;Halberstam, 2004;Hasanvand, Salehi, & Ebrahimipour, 2017;Heman-Ackah et al, 2003;Moers et al, 2012;Nuñez-Batalla et al, 2018;Phadke et al, 2018;Selamtzis, Castellana, Salvi, Carullo, & Astolfi, 2019). Even though CPP has been described as superior to HNR (Riesgo & Nöth, 2019), some works also show a strong and significant correspondence between HNR and 5 perceptual analysis of sustained vowels as a result of improvements in HNR estimation methods (Moers et al, 2012;Vaz Freitas et al, 2015).…”

“…For many years, vowel phonations have been favored over speech tasks in clinical measurements, due to its steady properties and higher relative reliability in the calculation of perturbation or spectralbased indices (Halberstam, 2004;Heman-Ackah et al, 2003;Heman-Ackah et al, 2002;Moers et al, 2012;Parsa & Jamieson, 2001;Zhang & Jiang, 2008). In more recent studies, the advantage of speech samples for its more natural communication context has been described, representing a great benefit to voice diagnostics (Delgado-Hernandez et al, 2018;Heman-Ackah et al, 2003;Heman-Ackah et al, 2014;Kitayama et al, 2018;Maryn & Weenink, 2015;Phadke et al, 2018;Sauder, Bretl, & Eadie, 2017;Selamtzis et al, 2019;Watts et al, 2017).…”

“…However, for CPP and CPPS the results regarding sustained vowels and connected speech were contradictory. While four studies found significant differences between healthy and dysphonic voices for both SV and speech samples (Brinca et al, 2014;Delgado-Hernandez et al, 2018;Hasanvand et al, 2017;Selamtzis et al, 2019), others reported higher CPP or CPPS values for SV as compared to speech (Kitayama et al, 2018;Phadke et al, 2018). Further, Gerrat et al (2016 reported no significant differences between SV and speech.…”

Premises for Effective Teaching and Learning: State of the Art, New Outcomes and Perspectives of Classroom Acoustics

“…On the basis of a study that investigated the voice quality measures used to discriminate different types of organic dysphonia from sustained [a:] vowels or [a:] vowels excerpted from speech, detected with microphone in the air in a quiet environment, the mean of the cepstral peak prominence smoothed (CPPS) distribution and the 95 th percentile of the sample entropy (SampEn) distribution showed better performance. 76 It also understood that when using CPPS or Sam-pEn there is an advantage of using the measures' distributions rather than their average values. Overall, it has been recognized that CPPS parameters obtained from distribution of occurrences obtained from both microphones in the air and contact microphones have a strong to good discrimination power related to an unhealthy voice.…”