Improved speaker recognition using DCT coefficients as features | IEEE Conference Publication | IEEE Xplore