Relationship of Cepstral Peak Prominence and Auditory-Perceptual Judgments of Severity in Varied Phonetic Contexts

Objective: Auditory perceptual evaluation of voice often includes multiple speech corpora developed to include varied phonetic contexts in an effort to elicit specific voice characteristics, such as hard onsets or easy onsets. These same speech corpora may be used for acoustic analysis of disordered voice (e.g., Analysis of Dysphonia in Speech and Voice; ADSV, Awan, 2011). However, clinicians may choose a smaller sample for acoustic analysis to save time. For example, clinicians may choose the all-voiced sentence from the Consensus Auditory-Perceptual Evaluation of Voice (CAPE-V; Kempster, et al., 2009), “We were away a year ago,” as the only sample of connected speech for acoustic analysis. This study sought to explore how Cepstral Peak Prominence (CPP), an acoustic measure of voice, elicited in various phonetic contexts correlates with an auditory-perceptual measure of overall severity. Further, this study sought to determine if particular phonetic contexts (e.g., the all-voiced sentence or the hard glottal attack sentence) are more predictive of an auditory-perceptual judgment for overall severity than others.

Methods: Cepstral Peak Prominence (CPP) was calculated for 296 samples from the Perceptual Voice Qualities Database (PVQD, Walden, 2022). The PVQD consists of voice samples collected using the CAPE-V sentences, providing the necessary varied phonetic contexts for this study. CPP values for each sentence and prolonged /a/ and /i/ were calculated individually. These measures were correlated with the average severity rating for overall severity provided by the PVQD. Further, CPP values for each phonetic context were used to predict the overall severity rating using regression analysis.

Results: Results provided Pearson Correlation Coefficients indicating the relative relationship between each sentence/prolonged vowel and the auditory-perceptual rating for overall severity. Regression analysis is ongoing.

Conclusions: This study provided information about the relative relationships observed between acoustic values resulting from differing phonetic contexts (i.e., CAPE-V speech stimuli) and auditory-perceptual judgments of overall dysphonia severity. These results may guide clinicians’ use of more efficient acoustic analyses to support clinical decision-making.