Objectives: Acoustic voice assessments are non-invasive and cost-effective for clinical use, but recording conditions—such as noise and reverberation—can impact the quality of acoustic analysis. The effect of room acoustics on voice assessment outcomes, however, is still not fully understood. This paper examines the impact of different levels of simulated reverberation on the accuracy and reliability of acoustic measurements for voice assessment, with the goal of identifying optimal recording conditions for clinical settings.

Methods: This study analyzed audio recordings of sustained /a:/ from 18 college-age English speakers (11 male and 7 female subjects). The recordings were corrupted with different reverberation effects using Audacity software, mimicking different room acoustic conditions. The reverberation time T20 of each reverberation effect was determined to characterize each simulated reverberation into different intensity levels. Using MATLAB and Praat software, we extracted five acoustic metrics including shimmer, jitter, smoothed cepstral peak prominence (CPPs), harmonics-to-noise ratio (HNR), and alpha ratio from both the clean and corrupted recordings. Statistical tests were conducted to identify significant variations in these measurements across various reverberation levels.

Results and Conclusions: The recordings convolved with different reverberation effects each defined by a specific reverberation time T20 indicated sensitivity of voice parameters to the reverberation. Analysis revealed that CPPs and jitter showed greater consistency to high levels of reverberation, with high T20 threshold values, whereas shimmer and HNR were more sensitive in reverberant environments, exhibiting lower T20 thresholds. Setting specific reverberation limits for voice clinics and recording spaces can enhance the precision of acoustic measurements, contributing to standardized room conditions for reliable voice assessment.