Integration of auxiliary features in Hidden Markov Models for Arabic speech recognition

Abstract

In this paper, the integration of auxiliary features in Hidden Markov Model (HMM) based Automatic Speech Recognition (ASR) system is presented. In particular, we concentrate on the potential benefits of the combination of auxiliary features with standard acoustic parameters in adverse acoustic environments. The experiments were fulfilled using the HTK Toolkit and ARADIGIT corpus which is a data base of Arabic spoken words. The obtained results show that while the integration of the auxiliary features with the standard parameters by SI (Separate Integration) strategy leads to small improvements in the two test environments (clean and noisy), their integration by DI (Direct Integration) strategy leads to a significant improvement of the recognition system performance in noisy environment.