基于随机共振的语音真实性检测方法  

Voice Authenticity Detection Method Based on Stochastic Resonance

作　　者：潘平[1] 罗辉[1] 李换双[1] 

机构地区：[1]贵州大学计算机科学与信息学院,贵阳550025

出　　处：《计算机工程》2013年第5期277-279,283,共4页Computer Engineering

基　　金：国家科技计划基金资助项目(2008RR0003);贵州省国际科技合作计划基金资助项目([2009]700109);贵州省科学技术基金资助项目([2012]2132)

摘　　要：已有的语音资料真实性检测方法是对内容或背景噪声进行的检测,导致对语音信号真实性检测效果不佳。为此,根据数字存储的本质特征和随机共振理论,提出一种语音信号真实性检测方法,从磁/荷的微弱变化中鉴别出编辑点。经大量的时域仿真实验,并与其他方法相比,证明该方法具有较高的鉴别率,并能判断出语音编辑点的位置。For now, some researches on the detection of voice authenticity are almost based on the recording material content and background noise, not involves the essence of recording material storage, resulting in poor authenticity detection results of the voice signal. According to the essence of digital storage and the stochastic resonance theory, a test method for voice signal authenticity is proposed. It can identify the edit points from the faint change of magnetism/charge. Many time domain simulation experiments show that this method has higher identification percentage than other methods, and can judge the location of the voice edit points.

关 键 词：数字存储 随机共振 语音 时域 真实性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]