Title :
Narrowband noise estimation in the subspace domain
Author :
Bhunjun, Vinesh ; Brookes, Mike
Author_Institution :
Commun. & Signal Process. Group, Imperial Coll., London, UK
Abstract :
In speech enhancement algorithms, the noise covariance matrix is normally estimated during silent periods detected using a VAD. In this paper an alternative method is proposed where the subspace complementary to the signal subspace is used to continuously estimate the noise covariance matrix. In addition, for narrowband noise an eigenvector matching algorithm is proposed to exploit interframe information for more reliable estimates.
Keywords :
covariance matrices; eigenvalues and eigenfunctions; parameter estimation; pattern matching; speech enhancement; eigenvector matching algorithm; interframe information; narrowband noise estimation; noise covariance matrix; speech enhancement algorithms; subspace domain; Acoustic noise; Colored noise; Covariance matrix; Eigenvalues and eigenfunctions; Frequency; Narrowband; Noise figure; Signal processing algorithms; Signal to noise ratio; Speech enhancement;
Conference_Titel :
Intelligent Multimedia, Video and Speech Processing, 2004. Proceedings of 2004 International Symposium on
Print_ISBN :
0-7803-8687-6
DOI :
10.1109/ISIMP.2004.1433985