Noise-Robust Method to Determine Speech Prosodic Characteristics to Assess Human Psycho-Emotional State in Free Motor Activity

摘要

The paper proposes noise-robust method to determine prosodic characteristics of speech. The nature of the method consists in decomposing a speech signal into informative noise and signal frequency components by means of improved complete ensemble empirical mode decomposition with adaptive noise, and selection of the component containing the pitch with the subsequent determination of prosodic characteristics. Decomposition used in the method does not need any a priori information (presetting) on the analyzed noisy speech signals. The study was conducted using a verified base of pure and noisy speech signals recorded in 220 males and females, aged 18 to 79 years, with signs of psycho-emotional disorders. The results show that the algorithm provides good noise robustness 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 for assessment of a human psycho-emotional state in free motor activity.