Blind Audio Source Separation Using Sparsity Based Criterion for Convolutive Mixture Case

Abstract

Abstract. In this paper, we are interested in the separation of audio sources from their instantaneous or convolutive mixtures. We propose a new separation method that exploits the sparsity of the audio signals via an p-norm based contrast function. A simple and efficient natural gradient technique is used for the optimization of the contrast function in an instantaneous mixture case. We extend this method to the convolutive mixture case, by exploiting the property of the Fourier transform. The resulting algorithm is shown to outperform existing techniques in terms of separation quality and computational cost.

Sparsity-based algorithms for blind separation of convolutive mixtures with application to EMG signals

“…It has been extended to the convolutive mixtures [2] by reformulating the data model in the timefrequency domain. Herein, we propose to use the ISBS for the separation of convolutive mixtures by rearranging them into instantaneous mixtures as done in (3).…”

“…Note Also that the use of sparsity can greatly improve the separation quality in overdetermined cases [1,2].…”

“…Two different methods are introduced for minimizing a sparsity-based cost function. The first technique is an extension to the convolutive mixtures of the ISBS algorithm presented in [1,2]. This algorithm has been used for the separation of instantaneous mixtures of audio signals based on their sparsity in the time domain [1] and the time-frequency domain [2].…”

“…The first technique is an extension to the convolutive mixtures of the ISBS algorithm presented in [1,2]. This algorithm has been used for the separation of instantaneous mixtures of audio signals based on their sparsity in the time domain [1] and the time-frequency domain [2]. The second technique proceeds to the minimization of the sparsity-based cost function using Givens rotations.…”