A tractable framework for estimating and combining spectral source models for audio source separation

• Published in: Signal processing Vol. 92; no. 8; pp. 1886 - 1901
• Main Authors: Arberet, Simon, Ozerov, Alexey, Bimbot, Frédéric, Gribonval, Rémi
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01.08.2012; Elsevier
• Subjects: Blind source separation; Blinds; Complexity; Computer Science; Convolutive mixture; Engineering Sciences; Estimating; Expectation–maximization algorithm; Gaussian mixture model; Mathematical models; Multichannel audio; Probabilistic methods; Probability theory; Separation; Signal and Image Processing; Spectra; Convolutive mixture; Blind source separation; Expectation–maximization algorithm; Multichannel audio; Gaussian mixture model; multichannel audio; expectation-maximization algorithm; convolutive mixture

The underdetermined blind audio source separation (BSS) problem is often addressed in the time–frequency (TF) domain assuming that each TF point is modeled as an independent random variable with sparse distribution. On the other hand, methods based on structured spectral model, such as the Spectral...