Title :
Instantaneous spectral estimation of nonstationary signals
Author :
Takizawa, Yumi ; Oda, Keisuke ; Fukasawa, Atsushi
Author_Institution :
Commun. Syst. Lab., Oki Electr. Ind. Co. Ltd., Tokyo, Japan
Abstract :
In this paper, new methods are proposed to estimate a nonstationary spectrum based on the covariance method (COV) and maximum entropy method (MEM). First, their evaluation functions are improved by an instantaneous expression. Secondly, the 2-D spectrum is defined as the spectrum frequency of the variation of the conventional spectrum, and band-limitation is achieved on the 2-D spectrum. The instantaneous covariance method (ICOV) and instantaneous maximum entropy method (IMEM) are newly derived from the above procedure. In this paper, the Wigner-Boashash method is also derived from the same viewpoint. IMEM was applied to analyze nonstationary signals of automobile vibration and speech signals
Keywords :
automobiles; covariance analysis; maximum entropy methods; parameter estimation; spectral analysis; speech processing; vibrations; 2-D spectrum; Wigner-Boashash method; automobile vibration; band-limitation; covariance method; evaluation functions; instantaneous covariance method; instantaneous maximum entropy method; instantaneous spectral estimation; maximum entropy method; nonstationary signals; spectrum frequency; speech signals; Autocorrelation; Automobiles; Engines; Entropy; Equations; Frequency; Prediction algorithms; Signal analysis; Spectral analysis; Speech analysis;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1994. ICASSP-94., 1994 IEEE International Conference on
Conference_Location :
Adelaide, SA
Print_ISBN :
0-7803-1775-0
DOI :
10.1109/ICASSP.1994.389740