Diagnosing clinical depression from voice: Using Signal Processing and Neural Network Algorithms to build a Mental Wellness Monitor

Untreated depression has been the major cause for suicide among youth. While psychotherapy, medication and other treatments have been successful in curing depression, majority of people suffering from depression do not actively seek, nor receive proper treatment. In many countries, mental health is not given as much importance as physical health and there is little or no awareness about mental disorders and their symptoms. Since depression has an effect on speech planning and production, we analyze a person's voice features to detect if they are depressed. We propose a model that processes the human voice signals given as input, converts it to a spectrogram image and builds a Neural Network Model that can be used to detect depression in humans. We implement this idea as a portable standalone device with a web interface as Proof of Concept.