Speaker adaptation of hybrid NN/HMM model for speech recognition based on singular value decomposition | IEEE Conference Publication | IEEE Xplore