Missing-feature-theory-based robust simultaneous speech recognition system with non-clean speech acoustic model | IEEE Conference Publication | IEEE Xplore