Title :
Noise-robust fundamental frequency extraction method based on band-limited amplitude spectrum
Author :
Shimamura, T. ; Takagi, H.
Author_Institution :
Dept. of Inf. & Comput. Sci., Saitama Univ., Japan
Abstract :
In this paper, a fundamental frequency extraction method of noisy speech is proposed. An exponentiated amplitude spectrum is calculated with a band-limitation operation and its inverse Fourier transform is used as the basis function for fundamental frequency extraction. It is shown that the function invokes a performance robust against noise as well as formant characteristics. Experimental results show that the proposed method provides an improvement relative to the conventional methods at a wide rage of signal-to-noise ratios.
Keywords :
Fourier transforms; signal denoising; speech enhancement; amplitude spectrum; fundamental frequency extraction; inverse Fourier transform; noisy speech; Additive noise; Autocorrelation; Data mining; Fourier transforms; Frequency; Noise level; Noise robustness; Noise shaping; Speech enhancement; Working environment noise;
Conference_Titel :
Circuits and Systems, 2004. MWSCAS '04. The 2004 47th Midwest Symposium on
Print_ISBN :
0-7803-8346-X
DOI :
10.1109/MWSCAS.2004.1354111