Performance evaluation of MLPC and MFCC for HMM based noisy speech recognition

In this paper auditory like features MLPC and MFCC have been used as front-end and their performance has been evaluated on Aurora-2 database for Hidden Markov Model (HMM) based noisy speech recognition. The clean data set is used for training and test set A is used to examine the performance. It has been found that almost the same recognition performance has been obtained both for MLPC and MFCC and the average word accuracy for MLPC and for MFCC is found to be 59.05% and 59.21%, respectively. It has also been observed that the MLPC is more effective than MFCC for noise type subway and exhibition, on the other hand, MFCC is more superior for babble and car noises.