MLLR method for Environmental Adaptation in a Continuous Farsi Speech Recognition

In this paper, MLLR adaptation of continuous density HMM is investigated in a Farsi speaker independent large vocabulary continuous speech recognition system in attempt to improve recognition rate in real world situations. In the MLLR framework, we have experienced the use of Gaussian mean transformations in global adaptation and regression tree based adaptation. Besides full and block-diagonal transformations of Gaussian means, transformation of Gaussian variances is examined. We have used MLLR technique in batch-supervised fashion since it is more beneficial in situations of severe mismatch. Our experiments on 4 different tasks, show that by using this technique the system recognition performance in a new environment can be significantly improved.