Implementing Vocal Tract Le in the Mllr Fram

Vocal Tract Length Normalization (VTLN) and Maximum Likelihood Linear Regression (MLLR) are two approaches to reduce the degradation in speech recognition performance caused by variation of speakers. This paper derives a novel efficient adaptation algorithm from the two techniques. Based on prior knowledge of usual VTLN, an approximate constrained-form linear transformation is obtained. The transformation is learned using EM algorithm and then applied in the MLLR setting. Experiments of three tasks are performed on an isolated word recognition system. Experimental results shows that with several adaptation words, WER is decreased greatly.