Abstract
This letter proposes a new method for concurrent voiced speech separation. Firstly the Wrapped Discrete Fourier Transform (WDFT) is used to decompose the harmonic spectra of the mixed speeches. Then the individual speech is reconstructed by using the sinusoidal speech model. By taking advantage of the non-uniform frequency resolution of WDFT, harmonic spectra parameters can be estimated and separated accurately. Experimental results on mixed vowels separation show that the proposed method can recover the original speeches effectively.
References
Ray Meddis, Lowel O’Mard, Psychophysically Faithful Methods for Extracting Pitch, Computational Auditory Scene Analysis, New Jersey, USA, Lawrence Erlbaum Associates, 1998, 43–58.
Zhao Heming, Zhu Meihong, Yu yibiao, Chen Xueqin, A multi-pitch detecting method suitable for CASA, Acta Sinica Eletronica, 31(2003)1, 123–126.
Huang Xiuxuan, Wei Gang, A new speech separation algorithm for auditory scene analysis, Journal of Electronics(China), 21(2004)3, 261–264.
C. O. Etemoglu, V. Cuperman, Matching pursuits sinusoidal speech coding, IEEE Trans. on Speech and Audio Processing, 11(2003)5, 413–424.
Stefan Franz, Sanjit K. Mitra, Gehard Doblinger, Frequency estimation using warped discrete Fourier transform, Signal Processing, 83(2003)8, 1661–1671.
Author information
Authors and Affiliations
Additional information
Supported by the National Natural Science Foundation of China (No.60172048).
Communication author: Zhang Xichun, born in 1963, male, Ph.D. student. School of Electronic and Info. Eng., South China Univ. of Tech., Guangzhou 510640, China.
About this article
Cite this article
Zhang, X., Li, Y., Zhang, J. et al. Concurrent speeches separation using Wrapped Discrete Fourier Transform. J. of Electron.(China) 22, 427–430 (2005). https://doi.org/10.1007/BF02687914
Received:
Revised:
Issue Date:
DOI: https://doi.org/10.1007/BF02687914