Stereo-based feature enhancement using dictionary learning

Abstract

This paper proposes stereo-based speech feature enhancement using dictionary learning. Instead of posterior values obtained by a Gaus sian mixture as in other methods, we use sparse weight vectors and their variants as an alternative noisy speech feature representation. This paper also provides an efficient algorithm that can be applied to large-scale speech processing. We show the effectiveness of the pro posed approach by using a middle vocabulary noisy speech recog nition task based on WSJ, which was provided by the 2nd CHiME Speech Separation and Recognition Challenge.Index Terms-Speech recognition, speech feature enhance ment, dictionary learning, sparse representation, 2nd CHiME chal lenge track 2