TR2007-002

Convolutive Speech Bases and their Application to Supervised Speech Separation
Citation: Smaragdis, P., "Convolutive Speech Bases and their Application to Supervised Speech Separation", IEEE Transaction on Audio, Speech and Language Processing, ISSN: 1558-7916, Vol. 15, Issue 1, pp. 1-12, January 2007 (IEEE Xplore)
Date:January 2007
MERL Contact:Bent Schmidt-Nielsen

In this paper we present a convolutive basis decomposition method and its application on simultaneous speakers separation from monophonic recordings. The model we propose is a convolutive version of the non-negative matrix factorization algorithm. Due to the non-negativity constraint this type of coding is very well suited for intuitively and efficiently representing magnitude spectra. We present results that reveal the nature of these basis functions and we introduce their utility in separating monophonic mixtures of known speakers.

 Read the full technical report (PDF: 322.9 kB)