Tree-structured model selection and simulated-data adaptation for environmental and speaker robust speech recognition | IEEE Conference Publication | IEEE Xplore