Pitch period estimation of voice signal based on EEMD and Hilbert Transform

An improved Empirical Mode Decomposition (EMD) named Ensemble Empirical Mode Decomposition (EEMD) is introduced to analyse voice signal. A brief review of EEMD and Hilbert Transform (HT) is given, and a flowchart of EMD/EEMD algorithm is proposed. With EEMD, a voice signal can be decomposed into a complete and finite set of adaptive basis functions defined as Intrinsic Mode Functions (IMFs), which admit well-behaved HT. Then the final presentation of the results is an energy-frequency-time distribution designated as the Hilbert Spectrum (HS), from which the pitch period of the voice signal can be perfectly estimated. The final experimental results show that HS plays a good performance and makes better results in the voice signal pitch period estimation.