摘要
语音识别是语音信号处理领域的研究热点，其训练效果的优劣直接关系到系统的整体性能。本文针对HMM和BP神经网络各自的优缺点，将两种方法有机地结合起来并应用于语音识别，进一步提高了语音识别的鲁棒性和准确率。
Speech recognition is a hot issue in the field of speech signal processing, the training unit is a very important part in speech recognition, and it determines the entire properties of the system. This article in view of HMM and the BP neural network respective good and bad points, organically unifies two methods and applies in the speech recognition, further enhanced the speech recognition robustness and the rate of accuracy.
出处
《跨世纪》
2008年第12期172-173,共2页
