Comparing multilayer perceptron to Deep Belief Network Tandem features for robust ASR

It is suggested that DBNs outperform single layer MLPs under the clean condition, but the gains diminish as the noise level is increased, while using MFCCs in conjunction with the posteriors from DBN's outperforms merely using single DBNS in low to moderate noise conditions.

Abstract

In this paper, we extend the work done on integrating multilayer perceptron (MLP) networks with HMM systems via the Tandem approach. In particular, we explore whether the use of Deep Belief Networks (DBN) adds any substantial gain over MLPs on the Aurora2 speech recognition task under mismatched noise conditions. Our findings suggest that DBNs outperform single layer MLPs under the clean condition, but the gains diminish as the noise level is increased. Furthermore, using MFCCs in conjunction with the posteriors from DBNs outperforms merely using single DBNs in low to moderate noise conditions. MFCCs, however, do not help for the high noise settings.