ABSTRACT
Psychiatric health is the most neglected facet of people around the world. Advances in information and technological growth created new avenues for public counselors and psychologists to gather data to have an in-depth understanding of psychological problems of the people. But psychiatric health still proves to be a challenging area for the human race. In this research, an attempt is made to build a novel model which can be used for preliminary screening to detect the onset of psychiatric problems (if any) and to predict and classify the status of psychiatric health using speech. Speech is the index of mind and the acoustic features of speech carry valuable information, which when analyzed can be used to predict the psychological wellbeing of a person. The model uses various kernels of support vector classifiers and applied on the speech data collected to optimize the classification and prediction of the psychiatric health at three different stages. The experiment was carried out on all the kernels. Results are compared and analyzed and an overall average accuracy of 93.2 percent was achieved.
