Pilot Study of Voice Biomarkers: Exploring Healthy Controls in a Non-Clinical Setting

Voice-based approaches for screening and diagnostic applications, particularly in telemedicine, often rely on patient recordings collected outside clinical environments. Establishing normative baselines is essential to advance voice analytics and clinical utility. This pilot study examined acoustic parameters in 32 healthy young adults (ages 18–24) with no history of vocal pathology, neurological disorders, or speech impediments. Participants provided paired recordings of sustained vowels (/a/, /e/, /o/, /u/) and a standardized phonetically balanced phrase (“The sun sets in Cincinnati on Saturday”). Analyses focused on features including fundamental frequency, jitter, shimmer, harmonics-to-noise ratio, formants (F1–F3), speaking rate, intensity, and spectral measures. Preliminary results revealed significant differences between healthy controls and a reference dataset of laryngitis patients, suggesting acoustic features can serve as objective markers of vocal fold inflammation. However, pathology-specific biomarker identification was constrained by the quality of available laryngitis data. Simple statistical comparisons proved insufficient, emphasizing the value of advanced measures such as cepstral peak prominence (CPP) and mel-frequency cepstral coefficients (MFCC). Challenges in non-clinical data collection highlight the need for standardized, detailed annotation of patient recordings to improve diagnostic accuracy and strengthen the predictive power of future biomarker studies.