Determining the Level of Depression using BDI-II through Voice Recognition

Abstract

Depression is a mental health disorder that is becoming a great threat to the mental well-being of every person experiencing it. To help those people in need by raising their awareness, the researchers developed a model of a system based on the questionnaire, Beck Depressive Inventory or BDIII, used by psychiatrists for assessing depression. The system is embedded on a small robot known by the name of Cozmo. The model of the system has the capability to determine the level of depression a person is experiencing through their voice features by utilizing two algorithms - Support Vector Machine (SVM) and Decision Tree (DT). Participant's speech was converted to text to obtain the level of depression by computing the score of their answers, while 34 voice features were extracted through every voice recording of each participant. A total of 84 participants, 42 college and 42 senior high school students participated in this study. The model achieved an accuracy rate of 40.5% for the SVM algorithm and 28.57% for the Decision Tree algorithm with both algorithms showing that the additional voice features that were added by the researchers with the initial voice features gained a higher accuracy.