EMD-Based Technique for Revealing Hidden Informative Speech Parameters to Assess Human Psycho-Emotional State

Abstract

The article presents a technique for revealing hidden informative parameters of speech based on the empirical mode decomposition, being an adaptive time-frequency analysis method. The proposed technique is based on the uniform splitting of the original speech signal into fragments, the empirical mode decomposition of fragments, and the formation of a set of informative (mode and composite) speech signals. Informative speech signals are formed due to the expansion of the informative space for amplitude-frequency, spectral-temporal and cepstral features of the original signal, being essential to identify hidden informative parameters. A synopsis of decomposition method types, along with pros and cons thereof have been demonstrated. The functionality of the proposed technique has been detailed, and the research outcomes have been reported. It has been concluded that the proposed technique actually provides an expansion of the informative space for the original speech signal features. An application of the technique for revealing hidden informative speech parameters will increase the efficiency for assessing human psycho-emotional state.