Improvement of the efficiency of voice control based on the complementary ensemble empirical mode decomposition

Low accuracy of voice recognition is one of the main problems of practical implementation of voice control systems (VCS). It is associated with the use of inefficient and non-adaptive methods for complex non-stationary voice signals processing. An improved algorithm to recognize voice commands using the adaptive processing technology of the Complementary Ensemble Empirical Mode Decomposition (CEEMD) is proposed. A block diagram and a detailed mathematical description of the algorithm basic blocks are given. The distinctive feature of the proposed algorithm is to extract only useful information on the unique properties of voice from the original speech signal. The experimental results of the proposed algorithm show the improving accuracy of voice commands recognition and efficiency of voice control as compared to the known RWTH ASR, Julius, and CMU Sphinx analogues.