DCT-based processing of dynamic features for robust speech recognition | IEEE Conference Publication | IEEE Xplore