A Directional Laplacian Density for Underdetermined Audio Source Separation

In this work, a novel probability distribution is proposed to model sparse directional data. The Directional Laplacian Distribution (DLD) is a hybrid between the linear Laplacian distribution and the von Mises distribution, proposed to model sparse directional data. The distribution's parameters are estimated using Maximum-Likelihood Estimation over a set of training data points. Mixtures of Directional Laplacian Distributions (MDLD) are also introduced in order to model multiple concentrations of sparse directional data. The author explores the application of the derived DLD mixtures to cluster sound sources that exist in an underdetermined two-sensor mixture. Key words: Audio Source Separation, Mixture Models, Directional Data, Sparse Data Modelling

• Type: Conference
• Year: 2010
• Where: ICANN
• Authors: Nikolaos Mitianoudis
• Keywords: Directional Laplacian Distributions, ICANN 2010, Linear Laplacian Distribution, Neural Networks, Sparse Directional Data