NONNEGATIVE MATRIX FACTORIZATION AND SPATIAL COVARIANCE MODEL FOR UNDER-DETERMINED REVERBERANT AUDIO SOURCE SEPARATION

DMCA

NONNEGATIVE MATRIX FACTORIZATION AND SPATIAL COVARIANCE MODEL FOR UNDER-DETERMINED REVERBERANT AUDIO SOURCE SEPARATION

Cached

Download Links

by Simon Arberet , Alexey Ozerov , Ngoc Q. K. Duong , Emmanuel Vincent , Rémi Gribonval , Frédéric Bimbot

Citations:

14 - 6 self

BibTeX

@MISC{Arberet_nonnegativematrix,
    author = {Simon Arberet and Alexey Ozerov and Ngoc Q. K. Duong and Emmanuel Vincent and Rémi Gribonval and Frédéric Bimbot},
    title = {NONNEGATIVE MATRIX FACTORIZATION AND SPATIAL COVARIANCE MODEL FOR UNDER-DETERMINED REVERBERANT AUDIO SOURCE SEPARATION},
    year = {}
}

OpenURL

Abstract

We address the problem of blind audio source separation in the under-determined and convolutive case. The contribution of each source to the mixture channels in the time-frequency domain is modeled by a zero-mean Gaussian random vector with a full rank covariance matrix composed of two terms: a variance which represents the spectral properties of the source and which is modeled by a nonnegative matrix factorization (NMF) model and another full rank covariance matrix which encodes the spatial properties of the source contribution in the mixture. We address the estimation of these parameters by maximizing the likelihood of the mixture using an expectation-maximization (EM) algorithm. Theoretical propositions are corroborated by experimental studies on stereo reverberant music mixtures. 1.

Keyphrases

full rank covariance matrix nonnegative matrix factorization theoretical proposition source contribution stereo reverberant music mixture convolutive case experimental study mixture channel time-frequency domain blind audio source separation spectral property zero-mean gaussian random vector spatial property

DMCA