中文摘要：

针对现代通信系统中压缩语音信号的特点,给出了编码域说话人识别的框架.在此框架下,研究了从GSM FR、G.729、G.723.1（6.3 k）3种语音编码器的码流中直接提取参数的方法.实验表明,在识别准确率和计算速度上,编码域说话人识别均优于传统的基于解码语音的说话人识别方法.

英文摘要：

According to the character of the modern communication system, we propose a new framework of speaker recognition in speech coding domain. In the framework, we study on feature extracting from GSM FR,G. 729 ,G. 723.1 （6.3 k） bit stream. Experiments show that the recognizing performance and efficiency of speaker recognition in speech coding domain are both better than the traditional method.