EMD-Based Voiced Speech Processing Method for Intelligent Recognition Systems of Stressed States in Humans

A method for voiced speech processing to be used in intelligent systems for recognizing human stressed states is proposed. Voiced speech fully reflects the violation in functioning of organs of human speech apparatus due to stressful emotional excitation of a person. The key feature of the method is to decompose voiced speech using the improved complete ensemble empirical mode decomposition with adaptive noise, to detect modes containing periodic information on the excitation source of the vocal tract, and to construct a composite signal reflecting information on glottal activity. The research results of the method, being computation of the pitch frequency of 100 multi-harmonic signals, with modulation in the frequency range of 0-2.5 Hz/ms in 0.5 Hz/ms increments, are presented. The modulation simulated the irregularity in vibration of the vocal folds (30-40 % of the nominal value) arising due to stressful emotional excitation. In accordance with the research results, it was concluded that the proposed method for voiced speech processing, using the improved decomposition, can be successfully tested in intelligent systems for recognizing stressed states in humans.