Integrating Depression Detection System: Face Speech and Text Analysis using Machine Learning

The increasing prevalence of depression in modern society has spurred research efforts aimed at developing automated, scalable, and non-invasive methods for its early detection. Among these methods, those based on speech analysis hold promise, yet there remains a need for techniques that can effectively capture articulatory information across various recording devices and in real-world settings. In this paper depression detection using Integrated Face, text and speech is analyzed. The proposed work integrates features of speech, image, and text modalities for accurate prediction of depression. Through integrated mechanisms, the overall learning efficiency is improved for accurate decision making. Extensive experimentation is conducted to build regression models for audio, video, and text modalities. Evaluation on datasets such as DAIC-WOZ and SH2 demonstrates the effectiveness of the proposed approach is compared with existing land mark features and its fusion.