Adaptive step size independent vector analysis for blind source separation

Abstract

In this paper, a novel adaptive step size independent vector analysis (ASS-IVA) method is proposed for blind source separation. Independent vector analysis (IVA) can successfully solve the classical permutation problem in the blind source separation (BSS) field. In the ASS-IVA method the step size is adjusted during learning to enhance the convergence behavior of the conventional IVA algorithm. The experimental results confirm that the proposed method improves the convergence speed greatly as compared to the original IVA method, whilst retaining the excellent separation properties of the IVA method.