Voice Quality in Multiple Sclerosis: A Clinical Perspective on Cepstral Peak Prominence-Smoothed and Harmonic-to-Noise Ratio

Objectives/hypothesis: This study aims to evaluate voice quality in people with multiple sclerosis (pwMS) with moderate-to-severe disability using objective acoustic measures: Cepstral peak prominence-smoothed (CPPS) and harmonic-to-noise ratio (HNR). The objectives are to compare these measures between pwMS and healthy subjects (HS) and assess their diagnostic accuracy in identifying inadequate voice quality and the need for voice rehabilitation. We hypothesize that pwMS will exhibit reduced voice quality compared to HS, with CPPS being a more reliable marker than HNR for assessing vocal impairments and guiding rehabilitation decisions. Study design: This cross-sectional study involved 34 pwMS with moderate-to-severe disability and 40 sex-matched HS. Methods: Participants’ voice recordings were collected through a 1-minute monologue and analyzed using PRAAT software (University of Amsterdam, Amsterdam, The Netherlands). Acoustic measures, CPPS and HNR, were extracted and compared between groups. Clinical perceptual judgment was employed as the gold standard. Diagnostic accuracy of CPPS and HNR was evaluated through receiver operating characteristic curve analysis. Results: CPPS and HNR values were significantly lower in pwMS compared to HS, indicating impaired voice quality due to reduced harmonicity and increased aperiodicity. CPPS (threshold = 11.07 dB) demonstrated high sensitivity (0.76) and specificity (0.82) for detecting inadequate voice quality. HNR (threshold = 11.29 dB) exhibited lower sensitivity (0.73) and specificity (0.73). CPPS proved to be a more reliable measure than HNR for identifying vocal impairments and determining the need for rehabilitation. Conclusions: Our findings confirm significant voice alterations in pwMS compared with a HS group in all collected measures. Although the data showed that the CPPS can be a valuable tool for voice assessment in multiple sclerosis, further studies should explore its applicability in broader populations and its integration into clinical protocols for voice assessment in pwMS.