I-Vector Based Depression Level Estimation Technique

Abstract

Depression is considered as a psychosomatic state

associated with the soft biometric features. People suffering from

depression always behave abnormal. Depression is a clinically

proven disorder that can overwhelm a person and his ability to

perform even a simple task. Soft biometric provides important

information about a person without being enough for their

verification because they lack uniqueness. This statement

comprises of features which are associated with the

psychosomatic state of a person such as feelings, sentiments or

brain related disorders like depression. In this thesis we have

estimated the depression level of each speech signal using I-

Vector technique. In our proposed approach first of all we have

removed silence from the speech signal then we have extracted

features from audio using I-Vector after that split overlapping

function is applied to evaluate overlapped audio beats. In the end

we have evaluated depression using relationship matrix. We have

estimated the depression level of each speaker. This technique

has better performance as compared with existing techniques.

The overall result has shown that the I-Vector technique has

good accuracy to detect depression in audios.