The relationships among physiological, acoustical, and perceptual measures of vocal effort

• Main Author: McKenna, Victoria
• Other Authors: Stepp, Cara E.
• Language: en_US
• Published: 2018
• Subjects: Speech therapy
• Online Access: https://hdl.handle.net/2144/30962

The purpose of this work was to explore the physiological mechanisms of vocal effort, the acoustical manifestation of vocal effort, and the perceptual interpretation of vocal effort by speakers and listeners. The first study evaluated four proposed mechanisms of vocal effort specific to the larynx: intrinsic laryngeal tension, extrinsic laryngeal tension, supraglottal compression, and subglottal pressure. Twenty-six healthy adults produced modulations of vocal effort (mild, moderate, maximal) and rate (slow, typical, fast), followed by self-ratings of vocal effort on a visual analog scale. Ten physiological measures across the four hypothesized mechanisms were captured via high-speed flexible laryngoscopy, surface electromyography, and neck-surface accelerometry. A mixed-effects backward stepwise regression analysis revealed that estimated subglottal pressure, mediolateral supraglottal compression, and a normalized percent activation of extrinsic suprahyoid muscles significantly increased as ratings of vocal effort increased (R2 = .60). The second study had twenty inexperienced listeners rate vocal effort on the speech recordings from the first study (typical, mild, moderate, and maximal effort) via a visual sort-and-rate method. A set of acoustical measures were calculated, including amplitude-, time-, spectral-, and cepstral-based measures. Two separate mixed-effects regression models determined the relationship between the acoustical predictors and speaker and listener ratings. Results indicated that mean sound pressure level, low-to-high spectral ratio, and harmonic-to-noise ratio significantly predicted speaker and listener ratings. Mean fundamental frequency (measured as change in semitones from typical productions) and relative fundamental frequency offset cycle 10 were also significant predictors of listener ratings. The acoustical predictors accounted for 72% and 82% of the variance in speaker and listener ratings, respectively. Speaker and listener ratings were also highly correlated (average r = .86). From these two studies, we determined that vocal effort is a complex physiological process that is mediated by changes in laryngeal configuration and subglottal pressure. The self-perception of vocal effort is related to the acoustical properties underlying these physiological changes. Listeners appear to rely on the same acoustical manifestations as speakers, yet incorporate additional time-based acoustical cues during perceptual judgments. Future work should explore the physiological, acoustical, and perceptual measures identified here in speakers with voice disorders. === 2019-07-06T00:00:00Z