摘要
为了提高语音识别系统的性能，基于全域优化的思想，提出了一种用于训练连续隐马尔柯夫模型（ＣＨＭＭ）的新算法——基因算法，并将该算法用于语音识别．用该算法训练ＣＨＭＭ，可得到最佳的模型参数，从而提高了语音识别率．利用该算法训练ＣＨＭＭ，不需要对ＣＨＭＭ的每一个参数单独进行估值，能够在一定的程度上提高训练速度．文中阐述了整个算法，给出了计算机模拟结果，并与传统的训练方法进行了比较．
For enhancing performance of speech recognition systems, a new algorithm based on global optimization about training CHMM——genetic algorithm is proposed, which is used for speech recognition. As the optimal model can be obtained by this method, the recognition rate is improved. The genetic algorithm evaluates all parameters in every training program, which can enhance the training speed. The whole algorithm is described in detail. The results of computer simulation and a comparison with the classical algorithm are given.
出处
《上海交通大学学报》
EI
CAS
CSCD
北大核心
1998年第6期19-22,共4页
Journal of Shanghai Jiaotong University
基金
国家自然科学基金
关键词
隐马尔柯夫模型
基因算法
语音识别
continuous hidden Markov model
genetic algorithm
speech recognition
