论文部分内容阅读
对窗口神经网络（ＢＷＮＮ）与鲁棒ＨＭＭ’ｓ之间的关系作了新的探讨，指出了鲁棒ＨＭＭ’ｓ是循环的ＢＷＮＮ（ＲＢＷＮＮ）的特殊情况。讨论了由ＲＢＷＮＮ构造用于噪音环境下语音识别的组合ＨＭＭ’ｓ和二阶ＨＭＭ’ｓ的算法，给出了它们相应的网络结构。
The relationship between window neural network (BWNN) and robust HMM’s is discussed. It is pointed out that the robust HMM’s is a special case of circular BWNN (RBWNN). The algorithm of constructing combined HMM’s and second-order HMM’s using RBWNN for speech recognition under noisy environment is discussed, and their corresponding network structures are given.