基于小波包的回放语音检测算法  

Playback Speech Detection Algorithm Based on Wavelet Packet

作　　者：汤爽 张二华[1] 唐振民[1] TANG Shuang;ZHANG Erhua;TANG Zhenmin(School of Computer Science and Engineering,Nanjing University of Science and Technology,Nanjing 210094)

机构地区：[1]南京理工大学计算机科学与工程学院,南京210094

出　　处：《计算机与数字工程》2022年第2期238-242,441,共6页Computer & Digital Engineering

基　　金：南京理工大学社会公共安全科技协同创新中心开放项目(编号:KY201706)资助。

摘　　要：以便携式回放设备的语音为代表的假冒语音攻击,给说话人识别系统带来了严峻的挑战。针对这种回放语音攻击问题,论文提出一种基于小波包的多频带回放语音鉴别算法。首先,通过小波包分解及重构后的信号进行傅里叶变换,取每一帧频谱的最大值;然后,利用对数运算以及离散余弦变换(DCT)来得到鉴别特征;最后,使用高斯混合模型(GMM)作为分类器进行假冒语音判别。实验表明,该检测算法能有效地鉴别回放语音。The fake voice attack represented by the voice of portable playback devices has brought severe challenges to the speaker recognition system. Aiming at this kind of playback voice attack problem,this paper proposes a multi-band playback speech discrimination algorithm based on wavelet packets. Firstly,the Fourier transform is performed on the signal after wavelet packet decomposition and reconstruction,and the maximum value of the spectrum of each frame is taken,Then,the logarithmic operation and discrete cosine transform(DCT)are used to obtain the discriminative features. Finally,the Gaussian mixture model(GMM)is used as a classifier to distinguish fake speech. Experiments show that the detection algorithm can effectively discriminate playback speech.

关 键 词：小波包分解 回放语音检测 高斯混合模型 说话人识别 

分 类 号：TP301.6[自动化与计算机技术—计算机系统结构]