From Screening to Precision: Searching for Voice Disorder-Specific Acoustic and Auditory-Perceptual Metrics

Area of Science:

• Laryngology and Speech Science
• Computational Acoustics
• Medical Diagnostics

Background:

• Acoustic and auditory-perceptual voice analysis are standard for screening voice disorders.
• Disorder-specific acoustic signatures for differential diagnosis are underutilized.
• Robustness of acoustic patterns across speech materials needs clarification for clinical applications.

Purpose of the Study:

• Investigate disorder-specific acoustic metrics for voice pathologies.
• Assess the consistency of acoustic patterns across different speech materials.
• Explore multidimensional voice quality patterns for differential diagnosis.

Main Methods:

• Utilized the Perceptual Voice Qualities Database.
• Employed Generalized Linear Models to associate acoustic parameters with pathologies (VFP, Atrophy, Lesions, MTD).
• Applied Principal Component Analysis (PCA) and Receiver Operating Characteristic (ROC) curves for pattern identification and performance evaluation.

Main Results:

• Two principal components (PC1, PC2) captured voice quality and stability.
• Vocal Fold Paralysis (VFP) showed strong discriminative performance (AUC ≥ 0.75).
• Atrophy, Lesions, and Muscle Tension Dysphonia (MTD) had moderate associations (AUC = 0.52-0.66); patterns were consistent across sustained vowels and connected speech.

Conclusions:

• Distinct acoustic-perceptual signatures for voice pathologies can be identified via multidimensional analysis.
• Findings support precision-based voice assessment and disorder-specific diagnostics.
• Robustness across speech materials facilitates flexible protocols and enhanced diagnostic/ML tools.