Effective use of DCTS for contextualizing features for speaker recognition | IEEE Conference Publication | IEEE Xplore