Speech enhancement based on Hilbert-Huang transform

The newly developed Hilbert-Huang transform (HHT) is introduced briefly in this paper. The HHT method is specially developed for analyzing nonlinear and non-stationary data. The method consists of two parts: (1) the empirical mode decomposition (EMD), and (2) the Hilbert spectral analysis. The EMD, which is the first part of the theory, can decompose any complicated data set into a finite and often small numbers of intrinsic mode functions (IMFs). IMFs also thus admit well-behaved Hilbert transforms. The law of its EMD and characteristics of the IMFs of a speech signal with unwanted sound are studied. Based on these studies, a new noise removal method has been developed. In the application, the HHT has been used to enhance the performance of speech signals by removing unwanted sound.