A method to determine cepstral markers of speech signals under psychogenic disorders

Abstract

This article proposes an automated method for detection of cepstral markers of speech signals under psychogenic disorders based on the method of improved complete ensemble empirical mode decomposition with adaptive noise. The method consists in decomposition of a speech signal with the help of the improved complete ensemble empirical mode decomposition with adaptive noise into frequency components with the subsequent formation of a set of informative components (concentration of information on psychogenic disorders), and the determination of their cepstral markers. A block diagram for the developed method is presented. The research is conducted using the generated verified signal base of healthy male and female patients, and patients with psychogenic disorders, aged 18 to 60 years. It follows in accordance with the results of the study that the psychogenic disorders affect the vocalized characteristics of the vocal tract, and are sufficiently fully displayed in cepstral markers. The proposed automated method can be used in remote monitoring systems of psychogenic disorders, and introduced into clinical practice of a psychiatrist to accelerate the treatment process.