说话人识别在智能家居中的应用  

Application of Speaker Recognition to Intelligent House

作　　者：龚伟[1] 

机构地区：[1]湖南师范大学工学院,湖南长沙410081

出　　处：《低压电器》2005年第9期22-24,32,共4页Low Voltage Apparatus

基　　金：湖南省教育厅资助项目(03C249)

摘　　要：说话人识别是当前语音信号处理的研究热点之一,它应用在身份认证领域具有得天独厚的优势。采用与文本相关的说话人识别方法,以Mel倒谱系数(Mel-Frequency Cepstral Coefficients,MFCC)及其一阶差分作为特征参数,采用矢量量化(VQ)技术进行模板创建。实验表明,该方法简便可靠,对于智能家居安防中的用户身份认证具有一定实用价值。Speaker Recognition is one of current researching hotspots of speech signal processing, which has many advantaged superiority at identity authentication field. The text-dependent speaker recognition method was presented, in which taking reel-frequency cepstral coefficients （MFCC） and its first order differentiation as characteristic parameters, adopting vector quantization （VQ） to build template. This simpler and reliable method and its certain practicability on identity authentication of safe-precaution for intelligent house were proved by experiments.

关 键 词：语言信号处理 特征参数 矢量量化 智能家居 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]