基于遗传算法的码本设计及其在说话人识别中的应用

矢量量化方法是与文本无关的说话人识别系统中广泛应用的建模方法之一.在矢量量化过程中,经典的LBG算法收敛速度快,但极易收敛于局部最优点,无法保证根据有限样本数据得到最优码本,并最终影响系统识别性能.本文提出了一种基于遗传算法并结合LBG算法的码本设计方法.该方法通过遗传操作获得全局优化的说话人模板,把新的识别算法应用于说话人识别的实验.结果显示,对于一个小的码本集,新的码本算法比经典的LBG算法具有更好的识别性能.

Codebook Design Using Genetic Algorithm and Its Application in Speaker Identification

Vector quantization is one of the popular modeling methods of codebook design in the text-independent speaker identification system at present.In the process of vector quantization,traditional LBG algorithm has the advantage of fast convergence,but it is easy to get the local optimal result, so the codebook designed by LBG is not surely optimal and the recognition performance will be influenced.In this paper,genetic algorithm(GA) is adopted to get global optimized speaker codebook and a new speaker identification algorithm based on GA algorithm and LBG is proposed.Experiment results show that this method improves the quality of codebook,the system based on GA performs with higher identification rates compared with that based on conventional LBG method.