Generalization of a maximum-likelihood approach to blind source separation | IEEE Conference Publication | IEEE Xplore