Speaker adaptation using discriminative linear regression on time-varying mean parameters in trended HMM | IEEE Journals & Magazine | IEEE Xplore