• 中国计算机学会会刊
  • 中国科技核心期刊
  • 中文核心期刊

J4 ›› 2011, Vol. 33 ›› Issue (4): 139-144.

• 论文 • 上一篇    下一篇

基于HMM和遗传神经网络的语音识别系统

包亚萍,郑〓骏,武晓光   

  1. (南京工业大学电子与信息工程学院,江苏 南京 210009)   
  • 收稿日期:2010-07-05 修回日期:2010-09-28 出版日期:2011-04-25 发布日期:2011-04-25
  • 基金资助:

    包亚萍(1965),女,江苏盐城人,硕士,副教授,研究方向为无线传感器网络、工业智能控制系统、信号处理和语音识别。郑骏(1986),男,江苏南京人,硕士生,研究方向为基于FPGA的信号处理、语音识别等。武晓光(1978),男,山西寿阳人,博士生,讲师,研究方向为图像目标识别与解析、数字信号处理、DSP技术等。

Speech Recognition Based on a Hybrid Model of Hidden Markov Models and the Genetic Algorithm Neural Network

BAO Yaping,ZHENG Jun,WU Xiaoguang   

  1. (School of Electronics and Information Engineering,Nanjing University of Technology,Nanjing 210009,China)
  • Received:2010-07-05 Revised:2010-09-28 Online:2011-04-25 Published:2011-04-25

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。

关键词: 语音识别, 隐马尔可夫模型(HMM), 遗传算法, 反向传播网络(BP)

Abstract:

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.

Key words: speech recognition;hidden markov models(HMM);genetic algorithm;BP neural networks(BP)