Multichannel Nonnegative Matrix Factorization in Convolutive Mixtures for Audio Source Separation

Abstract--We consider inference in a general data-driven object-based model of multichannel audio data, assumed generated as a possibly underdetermined convolutive mixture of source signals. We work in the short-time Fourier transform (STFT) domain, where convolution is routinely approximated as linear instantaneous mixing in each frequency band. Each source STFT is given a model inspired from nonnegative matrix factorization (NMF) with the Itakura

• Type: Journal
• Year: 2010
• Where: TASLP
• Authors: Alexey Ozerov, Cédric Févotte
• Keywords: Convolutive Mixtures, Data-driven Object-based Model, Short-time Fourier Transform, Software Engineering, TASLP 2010