Development of an SVM-based Depression Detection Model using MFCC Feature Extraction
V Maheshwar., N Venu Gopal., V. Naveen Kumar, Dharmala Pranavi, Y Padma Sai. · 2023
Early detection is crucial for effective therapy of depression, a mental disorder that impairs quality of life. With the accessibility of technology, writing articles about depression and identi fying those who could be depressed has become easier. In this paper, a machine learning-based speech-based depression detection system is built. The system uses natural language processing and voice analysis to extract information from speech patterns and determine if a person is depressed. It is difficult to create a model that matches all people bacause depression might seem different for each person and it can take a lot of time, money, and experienced specialists to properly collect and label data due to privacy and ethical issues, it might be challenging to obtain enough reliable data but our proposed system used DAIC-WOZ dataset as data source and used MFCC to extract relevant features and important characteristics from speech signals related to depression. The system is used to diagnose depression and has the potential for employing speech and language analytic methods to diagnose depression. Similar modes had achieved an accuracy between 75% to 80% but this paper achieved an accuracy of 89% which improves the functionality of the system compared to other systems.