Noise robust isolated word recognition

摘要

In this paper, we suggest a noise robust isolated word speech recognition system which can be applied in various noise environments. In this method, Kalman filter is used to remove the background noise and to enhance the speech signal. The enhanced signal is integrated into the front end of Dynamic Time Warping (DTW) isolated word recognition in order to guarantee high performance and robust recognition system. Performance of recognition word accuracy is evaluated under clean and different noisy environments.