Spectral Entropy in Speech for Classification of Depressed Speakers

This paper presents a study of spectral entropy analysis on speech for the possible prediction of depression in speakers who are at risk of committing suicide, when the symptom of depression strikes, unless admitted and having a proper treatment in time. Prediction is primarily necessary task to prevention of that life-threatening risk. In this study the full-band and further sub-band entropies of eight evenly separated frequency bands of 625 Hz estimated from the female voiced segments were computationally extracted and consequently used to form the parameter models for between-group classifications. The average of correct classification is considered to be fairly high when training a ML classifier with the 35% of extracted sample database and testing it again with the rest of sample database. As result shown, the classifying percentage obtained from study has suggested the higher frequency sub-band entropies extracted from spoken sound capable of being group discrimination between two categorized speakers.