基于声纹识别技术的常见模型与发展应用  

Common Models and Development Applications Based on VoiceprintRecognition Technology

作　　者：曾桂南 吴恋 何燕琴[1,2] 郭清粉 ZENG Guinan;WU Lian;HE Yanqin;GUO Qinfen(School of Mathematics and Big Data,Guizhou Education University,Guiyang 550018;Big Data Science and Intelligent Engineering Research Institute,Guizhou Education University,Guiyang 550018)

机构地区：[1]贵州师范学院,数学与大数据学院,贵阳550018 [2]贵州师范学院,大数据科学与智能工程研究院,贵阳550018

出　　处：《现代计算机》2021年第21期72-75,80,共5页Modern Computer

摘　　要：声纹识别也被称为说话人识别,是利用计算机技术及其他信息技术对声纹进行辨认从而对说话人的身份进行匹配的生物特征识别技术。声纹识别系统主要包括特征提取和模式识别,将主要以声纹识别技术的发展历程为主线,阐述常用于特征提取的线性预测倒谱系数,梅尔频率倒谱系数和几种常用于声纹识别系统的网络模型。Voiceprint recognition,also known as speaker recognition,is a biometric recognition technology that uses computer technology and other in⁃formation technologies to recognize voiceprints to match the identity of the speaker.The voiceprint recognition system mainly includes fea⁃ture extraction and pattern recognition.The development process of voiceprint recognition technology will be the main line,and the linear prediction cepstrum coefficients commonly used in feature extraction,Mel frequency cepstrum coefficients and several commonly used voiceprints will be explained.Identify the network model of the system.

关 键 词：声纹识别 发展 线性预测倒谱系数 梅尔频率倒谱系数 网络模型 

分 类 号：TN912.34[电子电信—通信与信息系统]