Text-indicated speaker recognition using kernel mutual subspace method | IEEE Conference Publication | IEEE Xplore