Sixth International Conference on Spoken Language Processing
(ICSLP 2000)

Beijing, China
October 16-20, 2000

A Computation-Efficient Parameter Adaptation Algorithm for the Generalized Spectral Subtraction Method

Jin-Jie Zhang, Zhi-Gang Cao, Zheng-Xin Ma

State Key Lab on Microwave and Digital Communications, Electronic Engineering Department, Tsinghua University, Beijing, China

The background noise in speech is not only objectionable to listeners with proper hearing but also very harmful to those hearing impaired. Hence, there is a strong need to employ speech enhancement to suppress the background noise in speech. In this paper, a computation-efficient parameter-adaptation algorithm is proposed for the generalized spectral subtraction method, which can reduce noise level more significantly.

Full Paper

Bibliographic reference.  Zhang, Jin-Jie / Cao, Zhi-Gang / Ma, Zheng-Xin (2000): "A computation-efficient parameter adaptation algorithm for the generalized spectral subtraction method", In ICSLP-2000, vol.4, 716-719.