中文摘要：

为了进一步提高语音识别系统的准确率,使语音产品应用更加方便,提出了一种隐马尔可夫模型和代数神经网络相结合的语音识别方法。利用隐马尔可夫模型生成最佳语音状态序列,将最佳状态序列的输出概率作为前馈型神经网络的输入,通过代数神经网络进行分类识别。使用Matlab7.0实验平台进行仿真,实验结果表明,与传统神经网络相比,该方法在收敛速度、鲁棒性和识别率方面都有改善。

英文摘要：

To improved the accuracy of speech recognition system,a new method for speech recognition is introduced,which combined the hidden Markov model（HMM） and the algebraic neural networks.Firstly,the HMM is used to generate the best speech state sequences,the output probabilities of the best speech state sequences is taken as the input of the feed-forward neural networks,then classify and recognize with algebra algorithm neural networks.Finally,the simulation result show the algorithm is better than the tradi-tional algorithm in convergence speed,robustness and recognition rate improvement.