A Method for Noise-Robust Speech Signal Processing to Assess Human Psycho-Emotional State

The nature of the noise-robust processing method consists in decomposing a speech signal into informative noise and signal frequency components using the improved complete ensemble empirical mode decomposition with adaptive noise, and selection of the component containing the pitch with the subsequent determination of frequency features. A simplified block diagram of the method with a brief mathematical description is presented. A study was conducted using a verified base of pure and noisy speech signals recorded from a group of subjects with signs of psycho-emotional disorders. In accordance with the results of the study, the method provides good robustness to noise of various intensities (signal-to-noise ratio is from 0 dB to 30 dB), and can be successfully tested in actual conditions of an "aggressive" noise environment to assess a person's psycho-emotional state.