论文部分内容阅读
模板匹配法技术是汉语声母识别中较为成功的算法,但它的缺陷影响了其恢复错误、改善识别性能。神经网络(NN)和模糊系统的结合,保留了双方的优点,充分利用了模糊神经网良好的容错性能、计算性能、分类性能和决策性能。本文重点研究了两种基于模糊神经网的声母识别方案,通过对其结构、识别率和特点的分析,可看出模糊神经网的声母识别性能明显优于模板匹配法,是更适于语音识别的网络。
The template matching method is a more successful algorithm in Chinese initial consonant recognition, but its defects affect its recovery error and improve the recognition performance. The combination of neural network (NN) and fuzzy system retains the advantages of both, and makes full use of the good fault-tolerant performance, computational performance, classification performance and decision-making performance of fuzzy neural network. This paper focuses on two kinds of initials recognition scheme based on fuzzy neural network. Through the analysis of its structure, recognition rate and characteristics, it can be seen that the performance of initial recognition of fuzzy neural network is better than that of template matching, which is more suitable for speech recognition network of.