Research of Speech Biomarkers for Stress Recognition Using Linear and Nonlinear Features

Psychological stress increases the risk of a number of mental disorders such as depression. Stress detection at early time is of great significance for the prevention of mental disorders. This speech-based research is to explore the effective biomarkers for stress detection. Linear and nonlinear features including pitch, short-term energy, sub-band energy ratio, teager energy operator (TEO), formant, mel-frequency cepstral coefficients (MFCC) were extracted to analyze speech under stress. The results showed that pitch, jitter and short-term energy variance can be effective biomarkers for stress recognition.