采用完整局部二进制模式的伪装语音检测

doi:10.11959/j.issn.1000-0801.2021108

Abstract

Abstract:

In view of the fact that the disguised speech detection algorithm based on local binary pattern (LBP) is not effective in detecting the spoofing attack from voice conversion, an anti-spoofing method based on completed local binary pattern (CLBP) was proposed.In this method, the spectrogram of speech signals is generated by the variable Q transformation (VQT) and used to train the true/spoofed speech classifier, so as to perform the detection of disguised speech.The experimental results demonstrate that the proposed anti-spoofing method based on the CLBP in the detection of voice conversion deception is better than the LBP-based algorithm, and when the parameter γ in VQT is set to 50, the detection system based on CLBP and SVM-RBF has the best performance for anti-spoofing the disguise speech.

Key words: speaker verification, anti-spoofing, CLBP, VQT

CLC Number:

TP391.42

Jian XU, Zhihua JIAN, Jiaqi YU, Yifan JIN, Lin YOU, Yunlu WANG. Completed local binary pattern based speech anti-spoofing[J]. Telecommunications Science, 2021, 37(5): 91-99.

Figures/Tables 9

References 18

[1]	吴震东, 潘树诚, 章坚武 . 基于CNN的连续语音说话人声纹识别[J]. 电信科学, 2017,33(3): 59-66.
	WU Z D , PAN S C , ZHANG J W . Continuous speech speaker recognition based on CNN[J]. Telecommunications Science, 2017,33(3): 59-66.
[2]	WU Z Z , DE LEON P L , DEMIROGLU C ,et al. Anti-spoofing for text-independent speaker verification: an initial database,comparison of countermeasures,and human performance[J]. IEEE/ACM Transactions on Audio,Speech and Language Processing, 2016,24(4): 768-783.
[3]	PHAPATANABURI K , WANG L B , NAKAGAWA S ,et al. Replay attack detection using linear prediction analysis-based relative phase features[J]. IEEE Access, 2019(7): 183614-183625.
[4]	林朗, 王让定, 严迪群 ,等. 基于逆梅尔对数频谱系数的回放语音检测算法[J]. 电信科学, 2018,34(5): 90-98.
	LIN L , WANG R D , YAN D Q ,et al. A playback speech detec-tion algorithm based on log inverse Mel-frequency spectral coefficient[J]. Telecommunications Science, 2018,34(5): 90-98.
[5]	KHODABAKHSH A , MOHAMMADI A , DEMIROGLU C . Spoofing voice verification systems with statistical speech synthesis using limited adaptation data[J]. Computer Speech ＆Language, 2017(42): 20-37.
[6]	WU Z Z , EVANS N , KINNUNEN T ,et al. Spoofing and countermeasures for speaker verification: a survey[J]. Speech Communication, 2015(66): 130-153.
[7]	PAUL D , PAL M , SAHA G . Spectral features for synthetic speech detection[J]. IEEE Journal of Selected Topics in Signal Processing, 2017,11(4): 605-617.
[8]	MUCKENHIRN H , KORSHUNOV P , MAGIMAI-DOSS M ,et al. Long-term spectral statistics for voice presentation attack detection[J]. IEEE/ACM Transactions on Audio,Speech and Language Processing, 2017,25(11): 2098-2111.
[9]	MOWLAEE P , SAEIDI R , STYLIANOU Y . Advances in phase-aware signal processing in speech communication[J]. Speech Communication, 2016(81): 1-29.
[10]	WANG L B , NAKAGAWA S , ZHANG Z F ,et al. Spoofing speech detection using modified relative phase information[J]. IEEE Journal of Selected Topics in Signal Processing, 2017,11(4): 660-670.
[11]	HANILCI C , . Speaker verification anti-spoofing using linear prediction residual phase features[C]// Proceedings of 2017 25th European Signal Processing Conference (EUSIPCO).[S.l.:s.n.], 2017.
[12]	OJALA T , PIETIKAINEN M , MAENPAA T . Multiresolution gray-scale and rotation invariant texture classification with local binary patterns[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2002,24(7): 971-987.
[13]	AHONEN T , HADID A , PIETIKAINEN M . Face description with local binary patterns: application to face recognition[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2006,28(12): 2037-2041.
[14]	ALEGRE F , VIPPERLA R , AMEHRAYE A ,et al. A new speaker verification spoofing countermeasure based on Local Binary Patterns[C]// Proceedings of 14th Annual Conference of the International Speech Communication Association (INTERSPEECH). Piscataway:IEEE Press, 2013.
[15]	GUO Z H , ZHANG L , ZHANG D . A completed modeling of local binary pattern operator for texture classification[J]. IEEE Transactions on Image Processing, 2010,19(6): 1657-1663.
[16]	CHANG C C , LIN C J . Libsvm[J]. ACM Transactions on Intelligent Systems and Technology, 2011,2(3): 1-27.
[17]	WU Z Z , YAMAGISHI J , KINNUNEN T ,et al. ASVspoof: the automatic speaker verification spoofing and countermeasures challenge[J]. IEEE Journal of Selected Topics in Signal Processing, 2017,11(4): 588-604.
[18]	徐剑 . 基于 CLBP 和 HOG 的伪装语音检测方法[D]. 杭州:杭州电子科技大学, 2020.
	XU J . A method for detection of spoofing speech based on CLBP and HOG[D]. Hangzhou:Hangzhou Dianzi University, 2020.

Metrics

Recommended 0

No Suggested Reading articles found!

子集	说话人		语音
子集	男生	女生	真实	伪装
训练集	10	15	3 750	12 625
开发集	15	20	3 497	49 875
评估集	20	26	9 404	184 000

子集		语音		声码器	伪装方式
子集	训练集	开发集	评估集	声码器	伪装方式
真实	3 750	3 497	9 404	无	无
S1	2 525	9 975	18 400	STRAIGHT	语音转换
S2	2 525	9 975	18 400	STRAIGHT	语音转换
S3	2 525	9 975	18 400	STRAIGHT	语音合成
S4	2 525	9 975	18 400	STRAIGHT	语音合成
S5	2 525	9 975	18 400	MLSA	语音转换
S6	0	0	18 400	STRAIGHT	语音转换
S7	0	0	18 400	STRAIGHT	语音转换
S8	0	0	18 400	STRAIGHT	语音转换
S9	0	0	18 400	STRAIGHT	语音转换
S10	0	0	18 400	STRAIGHT	语音合成

检测系统		γ =0			γ=25			γ=50
检测系统	已知伪装欺骗	未知伪装欺骗	平均值	已知伪装欺骗	未知伪装欺骗	平均值	已知伪装欺骗	未知伪装欺骗	平均值
LBP+GMM	5.381%	9.332%	7.357%	0.894%	4.350%	2.622%	0.477%	2.306%	1.392%
LBP+GMM-FA	5.053%	7.905%	6.479%	0.669%	3.873%	2.271%	0.451%	2.152%	1.302%
LBP+SVM-RBF	3.566%	9.830%	6.698%	0.482%	4.866%	2.674%	0.360%	2.270%	1.315%
CLBP_S/M/C+GMM	4.549%	8.025%	6.287%	0.369%	3.441%	1.905%	0.090%	1.502%	0.796%
CLBP_S/M/C+GMM-FA	3.934%	7.326%	5.630%	0.215%	2.508%	1.362%	0.065%	1.389%	0.727%
CLBP_S/M/C+SVM-RBF	3.206%	7.800%	5.503%	0.097%	2.630%	1.364%	0.029%	1.394%	0.712%

提取纹理特征的对象	提取纹理特征的方法			EER
提取纹理特征的对象	提取纹理特征的方法	S1	S2	S3	S4	S5	平均值
LFCC	LBP	1.318%	3.257%	0.860%	1.216%	2.442%	1.819%
	CLBP_S/M/C	1.059%	3.163%	0.507%	0.753%	1.962%	1.489%
MFCC	LBP	1.422%	5.869%	0.882%	1.304%	2.096%	2.315%
	CLBP_S/M/C	1.139%	5.291%	0.694%	0.873%	1.640%	1.927%
语谱图	LBP	0.690%	0.982%	0.150%	0.289%	0.747%	0.572%
	CLBP_S/M/C	0.248%	0.331%	0.000%	0.000%	0.318%	0.179%

提取纹理特征的对象	提取纹理特征的方法	已知的伪装欺骗						未知的伪装欺骗						所有伪装欺骗
提取纹理特征的对象	提取纹理特征的方法	S1	S2	S3	S4	S5	平均值	S6	S7	S8	S9	S10	平均值	平均值
LFCC	LBP	1.140%	2.951%	0.545%	0.940%	2.060%	1.527%	2.553%	1.364%	1.615%	1.253%	11.467%	3.650%	2.589%
	CLBP_S/M/C	0.896%	2.463%	0.315%	0.507%	1.431%	1.122%	1.758%	0.959%	0.914%	0.819%	10.339%	2.958%	2.040%
MFCC	LBP	1.029%	5.385%	0.529%	0.899%	1.968%	1.962%	4.857%	1.194%	1.231%	1.520%	17.720%	5.304%	3.633%
	CLBP_S/M/C	0.779%	4.890%	0.403%	0.492%	1.404%	1.594%	4.504%	0.785%	0.857%	0.900%	15.879%	4.585%	3.090%
语谱图	LBP	0.661%	0.796%	0.084%	0.145%	0.698%	0.477%	0.803%	0.752%	0.670%	0.596%	8.709%	2.306%	1.392%
	CLBP_S/M/C	0.125%	0.174%	0	0	0.153%	0.090%	0.190%	0.145%	0.103%	0.080%	6.991%	1.502%	0.796%

Completed local binary pattern based speech anti-spoofing

RichHTML

PDF下载

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 9

References 18

Related Articles 2

Metrics

Recommended 0

[1]	Honghui JIN, Zhihua JIAN, Man YANG, Chao WU. Synthetic speech detection method using texture feature based on circumferential local ternary pattern [J]. Telecommunications Science, 2023, 39(6): 85-95.
[2]	Jia XU, Zhihua JIAN, Honghui JIN, Chao WU, Lin YOU, Yingxiao WU. Synthetic spoofing speech detection method based on center-symmetric local binary pattern [J]. Telecommunications Science, 2023, 39(1): 72-78.