A unified method for blind separation of sparse sources with unknown source number
Qi Lv, Xian‐Da Zhang · IEEE Signal Processing Letters · 2005
Blind source separation (BSS) is of high interest in practical applications. Some approaches are proposed for the unknown number of sources of the BSS problem. However, they only consider the overdetermined case with the number of sensors more than the number of sources. To implement the practical BSS without prior assumption on the number of sources, we propose a new BSS method. It uses the unsupervised robust C prototypes algorithm to estimate the mixing matrix and then makes the estimation of sources. Simulations using speech signals confirm the validity of the proposed method.