Speech formant frequency estimation method based on Hilbert–Huang transform

Abstract

Translate Article

A speech formant frequency estimation method based on Hilbert–Huang transform (HHT) is proposed in this study. After filtering with bandpass filters with the center frequencies obtained by using the FFT analysis, speed data are decomposed into a set of intrinsic mode function (IMFs) by using the HHT analysis method. The IMFs containing formant frequencies are then identified according to the energy maximum criteria, their instantaneous frequencies and Hilbert spectra are calculated, and finally the formant frequencies of speech data are efficiently determined. The results in this study show that, compared with conventional formant estimation methods, the method based on HHT not only can give clearer descriptions of the nonlinear and nonstationary characteristics of speech signals, but also the speech formant frequencies and their variations with high time–frequency resolution and veracity. [This work was supported by Grant No. 60275004 from the National Natural Science Foundation of China (NNSFC).]