Speech enhancement using the bispectrum

Abstract
The authors present a study on the enhancement of speech that has been corrupted by additive wideband Gaussian noise. A least squares bispectrum reconstruction algorithm is used as a means for attenuating wideband Gaussian noise. A systematic approach is presented for processing speech records and estimating the bispectrum on a frame-by-frame basis. Performance evaluations are given using phonetically balanced sentences and signal-to-noise ratios.

This publication has 10 references indexed in Scilit: