Rapid feature space MLLR speaker adaptation for deep neural network acoustic modeling | IEEE Conference Publication | IEEE Xplore