人工蜂群算法改进DHMM的语音识别方法

摘要

针对离散隐马尔可夫(Discrete Hidden Markov Model,DHMM)语音识别系统中LBG算法对初始码书的依赖性和易陷入局部最优解的问题,采用人工蜂群(Artificial Bee Colony,ABC)算法对语音特征参数进行矢量量化,从而得到最优码书,提出了ABC改进DHMM的孤立词语音识别方法.先提取语音信号的特征参数,然后用ABC算法中每个食物源表示一个码书,以人工蜂群进化的方式对初始码书进行迭代而获得最优码书,最后把最优码书的码矢标号代入DHMM模型进行训练和识别.实验结果表明,ABC改进的DHMM语音识别方法与传统的LBG及粒子群优化初始码书的LBG的DHMM语音识别方法相比具有较高的识别率和较好的鲁棒性.%The paper proposes the modified DHMM speech recognition algorithm which uses Artificial Bee Colony algorithm (ABC) to cluster speech feature vector and generate the optimal codebook in the Discrete Hidden Markov Model (DHMM) speech recognition system. In the experiments, extract the feature vector of speech. In ABC algorithm, each food source indicates a codebook. The optimal codebook is obtained by using bee evolution ways to iterative initial codebook. The optimal codebook enters the DHMM to be trained and recognized. The experimental results show that the modified DHMM speech recognition algorithm has higher recognition ratio and better robustness than DHMM algorithm which uses the traditional LBG algorithm and the LBG algorithm of particle swarm optimization initial codebook.