Strategies for modeling reverberant speech in the feature domain | IEEE Conference Publication | IEEE Xplore