摘要
针对语音认证系统中高保真录音回放攻击问题,开展了录音回放攻击取证的研究,论述了基于信道模式噪声和回放语音MFCC特征参数提取的取证算法;通过labview软件编程进行仿真验证,提取了疑似回放攻击语音和原始语音信道噪声差值频谱图和疑似回放语音MFCC特征值,结果表明通过提取回放语音中信道模式噪声和MFCC特征参数并设计滤波器去噪能够识别原始语音和高保真录音的差异,为语音司法鉴定和语音认证系统的设计提供了理论依据。
Aiming at the playback attack in the speech certification system, In the study of the playback attack forensics, Model is presented based on the channel noise and playback voice MFCC feature parameter extraction of forensics algorithm , Through the labview software programming for simulation, Extract the suspected playback attack speech and original speech channel noise difference spectrum and suspected playback speech MFCC fea- ture values, Results show that by extracting playback speech mode noise in channel and MFCC feature parame- ters and design filter denoising to identify original speech differences and high-fidelity recording, For voice judi- cial authentication and voice provides theory basis for the design of the authentication system.
出处
《安阳工学院学报》
2015年第4期47-50,共4页
Journal of Anyang Institute of Technology
基金
安徽省高校省级优秀青年人才基金自然科学重点研究项目(编号:2013SQRL116ZD)
