Analysis of ElectroGlottoGraph signal using Ensemble Empirical Mode Decomposition

Abstract

The analysis of various components of the Electroglottograph (EGG) signal, obtained after Ensemble Empirical Mode Decomposition (EEMD) is the primary objective of this paper. The ability of EEMD to detect intermittent high frequency data embedded in the data of lower frequency is exploited to segregate the Epoch locations and the Periodic nature of EGG signal. The dyadic filterbank property of EEMD segregates the EGG signal into intrinsic mode functions (IMFs), in decreasing order of frequency. Hilbert envelope (HE) and moving average filter are used to determine the epoch locations and compute the pitch frequency from the first IMF, whereas pitch frequency is computed directly from latter IMFs. Block Processing of the EGG data is avoided and the results are evaluated with respect to the differential EGG (dEGG) signal.