Factored covariance modeling for text-independent speaker verification | IEEE Conference Publication | IEEE Xplore