Fifth International Workshop on Models and Analysis of Vocal Emissions for Biomedical Applications (MAVEBA 2007)

Florence, Italy
December 13-15, 2007

Predicting Severity of Mental State Using Vocal Output Characteristics

M. Landau (1), T. Yingthawornsuk (1), D. Mitchell Wilkes (1), Richard G. Shiavi (1,2), R. M. Salomon (3)

(1) Department of Electrical and Computer Engineering; (2) Department of Biomedical Engineering; (3) Department of Psychiatry, School of Medicine
Vanderbilt University, Nashville, TN, USA

Acoustical properties of speech have been shown to be related to mental states such as depression and remission. In particular, energy in frequency bands has been used as features for group classification among the groups with mental states of remission, depression, and suicidal risk. The prediction algorithms presented develop an additional level of assessment and are designed to predict a score for the severity of the mental state, provided by the Beck Depression Inventory. Several multiple regression models have been produced relating the results of the inventory and the power in four frequency bands. Models were produced for both males and females using both spontaneous and automatic speech.
Index Terms. speech, mental states, power spectra, regression

Full Paper (reprinted with permission from Firenze University Press)

Bibliographic reference.  Landau, M. / Yingthawornsuk, T. / Wilkes, D. Mitchell / Shiavi, Richard G. / Salomon, R. M. (2007): "Predicting severity of mental state using vocal output characteristics", In MAVEBA-2007, 153-156.