摘要
在说话人识别系统中,特征参数的选择和提取对系统的识别性能有关键性的影响。研究了两种重要的语音特征参数,线性预测倒谱系数和美尔倒谱系数,在此基础上提出改进的相位自相关系数,通过实验对几种参数进行了对比,结果表明改进的相位自相关系数能够使系统的误识率明显下降。
In speaker recognition system, the feature selection and extraction is one of the most important problems in speaker recognition. The two main acoustic feature parameters LPCC and MFCC were studied. The ameliorative phase auto correlation coefficient was proposed based on them. The experimentation results showed that the ameliorative phase auto correlation coefficient could obviously improve the recognition accuracy of the speaker recognition system.
出处
《大理学院学报(综合版)》
CAS
2009年第8期32-35,共4页
Journal of Dali University
关键词
说话人识别
特征参数
线性预测倒谱系数
美尔倒谱系数
speaker recognition
feature parameter
linear prediction eepstrum coefficients (LPCC)
reel-frequency cepstrum coefficients( MFCC )