基于HMM和遗传神经网络的语音识别系统

摘要

本文提出了一种基于隐马尔可夫(HMM)和遗传算法优化的反向传播网络(GA-BP)的混合模型语音识别方法.该方法首先利用HMM对语音信号进行时序建模,并计算出语音对HMM的输出概率的评分,将得到的概率评分作为优化后反向传播网络的输入,得到分类识别信息,最后根据混合模型的识别算法作出识别决策.通过Matlab软件对已有的样本数据进行训练和测试.仿真结果表明,由于设计充分利用了HMM时间建模能力强和GA-BP神经网络分类能力强等特点,该混合模型比单纯的HMM具有更强的抗噪性,克服了神经网络的局部最优问题,大大提高了识别的速度,明显改善了语音识别系统的性能.%A new method for speech recognition based on a hybrid model of hidden Markov models(HMM ) and the genetic algorithm neural network ( GA-BP ) is presented. The HMM is employed to compute the Viterbi output score. Then the score is used as the input of the GA-BP network to acquire the classification information. Finally, the sampled data are trained and tested by the Matlab software.And the result of recognition is made by the recognition information. The recognition experiment shows that the model has higher performance than the hidden Markov model in speech recognition, because of the dynamic time series, the greatly strengthened modeling ability of HMM, and the greatly strengthened classification ability of the GA-BP network.