72 Assessment of Depression from Speech

Spectral characteristics of speech signal have formerly been studied and reported for its quantitative information in relating the emotional state in persons suffering depression to the level of symptom affect. The spectral energy distributing in the particular bandwidths of audio frequency less than 5 KHz extracted from the subjects' speech sample reveals as the distinct parameter with a property that can represent the severity of symptom in term of energy shifting and amount change found in this study empirically. The acoustical analysis focused on Power Spectral Density (PSD) was performed on speech samples extracted from three different databases collected from the remitted, depressed, and suicidal patient groups. The result of classifications suggests the significant potential of PSD's in separating all categorized patient groups which consistently verifies that the psychomotor disturbance affects speech production system in speaker, clinically known as the central element in depression and other high-risk emotional disorders.