录音回放攻击检测技术综述  

Overview of replay attack detection technology

作　　者：周颖慧[1] 刘亚丽[1] ZHOU Yinghui;LIU Yali(Communication Acoustics Lab of Communication University of China,Beijing 100024,China)

机构地区：[1]中国传媒大学传播声学研究所

出　　处：《电声技术》2019年第8期76-80,共5页Audio Engineering

摘　　要：随着录音设备价格的降低,质量的提高,且不需要专业知识的优势,所以回放攻击已成为一种最常见、最重要的说话人识别系统攻击手段。本文介绍了录音回放的易破坏性,总结了现有关于各种攻击方式(模拟说话人攻击、语音合成攻击、语音转换攻击以及回放攻击)的相关研究,并详细对现有回放攻击检测技术进行了介绍与分析,为今后相关工作的开展提供了可靠全面的背景介绍。With the reduction of the price of recording equipment,the improvement of quality,and the lack of expertise,playback replay has become the most common and important speaker identification system attack.The article introduces the results of existing research work and proposes the key research directions in the future.It not only describes the vulnerability of recording playback and summarizes the existing research on various attack modes (simulated speaker attack,speech synthesis attack,voice conversion attack and playback attack),but also introduces and analysis the existing playback attack detection technology in detail.At last,it provides a reliable and comprehensive background for the development of related work in the future.

关 键 词：说话人识别 易破坏性 录音回放攻击 对抗方式 

分 类 号：TN912.34[电子电信—通信与信息系统]