矢量量化技术在说话人识别中的应用

Application of vector quantification in speaker recognition

矢量化(VQ)是一种及其重要的信号压缩方法,广泛应用于语音信号压缩等领域。说话人识别系统利用所建的VQ码本保存了说话人一定的声道特性,来进行说话人识别,完成这个识别系统有两个步骤:一是建立参考模型码本;二是进行模式匹配。

Vector quantification, widely used in areas like phonetic signal compressing, is one of the important methods of signal compressing. The speaker recognition system retains the sound track characteristics of certain speakers, using code vector, and thus achieves the purpose of recognizing particular speaker. It will have to undergo two steps to construct this recognition system： establishing consulting model code vector, matching the mode.