Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

Robust Speaker Recognition Based on High Order Cumulant

Jiyong Ma, Wen Gao

Institute of Computing Technology,, Chinese Academy of Sciences, Beijing, China

LP-derived cepstral coefficients are sensitive to additive noise in speech signal. In this paper, an approach to extracting speech feature based on the high-order cumulant is proposed to depress the effect of additive noise in speech signal. The performance of this approach is evaluated using a text-prompt speaker verification system. Experimental results show that this approach is effective to increase the robustness of the speaker recognition system.


Full Paper

Bibliographic reference.  Ma, Jiyong / Gao, Wen (2000): "Robust speaker recognition based on high order cumulant", In ICSLP-2000, vol.2, 278-281.