Noise robust speech recognition by integration of MLLR adaptation and feature extraction for noise reduced speech

摘要

In this paper, we investigate a noise robust acoustic feature in our proposed noise robust speech recognition method using Kalman filtering for speech signal estimation and iterative unsupervised MLLR adaptation. For the noise robust acoustic feature, we employed root cepstral coefficients and compared the results with conventionally used MFCCs at speech recognition accuracy. Furthermore, we investigate the number of phoneme clusters in MLLR adaptation. In order to evaluate the proposed method, we carried out large vocabulary continuous speech recognition experiments under 3 types of music. As a result, the proposed method showed the significant improvement in word accuracy.