Acoustic differences between healthy and depressed people: a cross-situation study

Area of Science:

• Psychiatry
• Speech Science
• Computational Linguistics

Background:

• Vocal abnormalities are common in depression but their situational specificity is unclear.
• Previous studies often failed to control for demographic variables influencing voice.
• This study investigates vocal differences in depression across various situations, controlling for covariates.

Purpose of the Study:

• To determine if vocal abnormalities in depression are situation-specific.
• To compare vocal characteristics between depressed and healthy individuals in 12 distinct speech scenarios.
• To identify acoustic features that differentiate between depressed and healthy individuals.

Main Methods:

• Comparison of vocal differences between depressed patients and healthy controls in 12 speech scenarios.
• Analysis of positive, negative, and neutral voice expressions across four tasks.
• Utilized Multiple Analysis of Covariance (MANCOVA) to assess group effects on acoustic features, considering effect size.

Main Results:

• Significant vocal differences were found between depressed and healthy groups in all 12 speech scenarios.
• While specific acoustic features varied by scenario, loudness, MFCC5, and MFCC7 showed consistent differences with large effect sizes.
• These three features were consistently distinct between the two groups.

Conclusions:

• Vocal differences between depressed and healthy individuals are present across 12 evaluated scenarios.
• Loudness, MFCC5, and MFCC7 are potential acoustic indicators for depression detection through voice analysis.
• Depression influences voice through both situation-specific and cross-situational acoustic patterns.