摘要
提出一种结合自组织特征映射（Ｓｅｌｆ－ｏｒｇａｎｉｚｉｎｇＦｅａ－ｔｕｒｅＭａｐ，ＳＯＭ）和多层感知器（ＭｕｌｔｉｌａｙｅｒＰｅｒｃｅｐｔｒｏｎ，ＭＬＰ）的神经网络语音识别系统。
A new speech recognition system which combines SOM (Self organizing Feature Map) with MLP (Multilayer Perceptron) neural networks for isolated word recognition is presented. Experimental results show that the recognition performance of the new system is much better than that of the conventional MLP recognizer. The new system is expected to use in more words data base.
出处
《数据采集与处理》
CSCD
1996年第2期119-122,共4页
Journal of Data Acquisition and Processing
基金
国家攀登计划认知科学(神经网络)重大关键项目
