ESTIMATION OF THE PITCH FREQUENCY SPEECH SIGNALS METHODS                                     EMPIRICAL MODE DECOMPOSITION

Alimuradov Alan Kazanferovich, applicant, sub-department of information and measuring equipment and metrology, researcher, research department, Penza State University, alansapfir@yandex.ru

004.934

In order to enhance the evaluation of pitch frequency estimation in the preprocessing a new method of adaptive decomposition of nonlinear and non-stationary speech signals – a method of empirical mode decomposition (EMD). Based on our review provides a detailed mathematical description of the types of methods of decomposition (Ensemble Empirical Mode Decomposition EEMD, Complementary Ensemble Empirical Mode Decomposition CEEMD). The investigations of the possibility of EMD, and EEMD CEEMD in algorithm of the pitch estimation, based on analysis of the energy and spectral characteristics of the empirical modes. In accordance with the results it is concluded that the most adaptive decomposition method is a method of speech signals CEEMD which may find practical application in algorithms of the pitch frequency estimation.

speech signal processing, Empirical Mode Decomposition, Ensemble Empirical Mode Decomposition, Complementary Ensemble Empirical Mode Decomposition, pitch frequency.