基于设备本底噪声频谱特征的手机来源识别
|
摘要
随着手机录音设备的普及以及各种功能强大且易于操作的数字媒体编辑软件的出现,手机来源识别已成为多媒体取证领域的热点问题。将本底噪声作为手机的"指纹",提出了一种基于本底噪声的手机来源识别方法。该方法先通过静音段录音的估计得到本底噪声;然后计算本底噪声的频谱特征在时间轴方向上的均值,将其作为手机来源识别的分类特征;最后采用主成分分析(PCA)法对特征进行降维,并采用支持向量机(SVM)进行分类。实验部分对24款主流型号的手机进行了分类,结果表明本文方法的平均识别准确率(accuracy)和平均召回率(recall)达到了99.24%,同时也验证了相比MFCC,本底噪声有更加优越的性能。
With the widespread availability of cell-phone recording devices and the availability of various powerful and easy-to-use digital media editing software, source cell-phone identification has become a hot topic in multimedia forensics. A novel cell-phone identification method was proposed based on the recorded speech. Firstly, device self-noise(DSN) was considered as the fingerprint of the cell-phone and estimated from the silent segments of the speech. Then, the mean of the noise's spectrum was extracted as the identification. Principal components analysis(PCA) was applied to reduce the feature dimension. Support vector machine(SVM) was adopted as the classifier to determine the source of the detecting speech. Twenty-four popular models of the cell-phones were evaluated in the experiment. The experimental results show that the average identification accuracy and recall of the method can reach up to 99.24% and demonstrate that the self-noise feature has more superior performance than the MFCC feature.
With the widespread availability of cell-phone recording devices and the availability of various powerful and easy-to-use digital media editing software, source cell-phone identification has become a hot topic in multimedia forensics. A novel cell-phone identification method was proposed based on the recorded speech. Firstly, device self-noise(DSN) was considered as the fingerprint of the cell-phone and estimated from the silent segments of the speech. Then, the mean of the noise's spectrum was extracted as the identification. Principal components analysis(PCA) was applied to reduce the feature dimension. Support vector machine(SVM) was adopted as the classifier to determine the source of the detecting speech. Twenty-four popular models of the cell-phones were evaluated in the experiment. The experimental results show that the average identification accuracy and recall of the method can reach up to 99.24% and demonstrate that the self-noise feature has more superior performance than the MFCC feature.
引文
[1]杨锐,骆伟祺,黄继武.多媒体取证[J].中国科学:信息科学,2013,43(12):1654-1672.YANG R,LUO W Q,HUANG J W.Multimedia forensics[J].Science China:Information Science,2013,43(12):1654-1672.
[2]GUPTA S,CHO S,KUO C,et al.Current developments and future trends in audio authentication[J].IEEE Multimedia,2012,19(1):50-59.
[3]BAO Y,LIANG R.Research progress on key technologies of audio forensics[J].Journal of Data Acquisition&Processing,2016,31(2):252-259.
[4]王昌海,张建忠,徐敬东,等.基于HMM的动作识别结果可信度计算方法[J].通信学报,2016,37(5):143-151.WANG C H,ZHANG J Z,XU J D,et al.Identifying the confidence level of activity recognition via HMM[J].Journal on Communications,2016,37(5):143-151.
[5]贺前华,王志锋,RUDNICKYA I,等.基于改进PNCC特征和两步区分性训练的录音设备识别方法[J].电子学报,2014,42(1):191-198.HE Q H,WANG Z F,RUDNICKYA I,et al.A recording device identification algorithm based on improved PNCC feature and two-step discriminative training[J].Acta Electronica Sinica,2014,42(1):191-198.
[6]HANILCI C,ERTAS F,ERTAS T,et al.Recognition of brand and models of cell-phones from recorded speech signals[J].IEEE Transactions on Information Forensics&Security,2012,7(2):625-634.
[7]HANILCI C,CEMAL I,ERTAS F.Optimizing acoustic features for source cell-phone recognition using speech signals[C]//ACM Workshop on Information Hiding and Multimedia Security,June20-22,2013,Montpellier,France.New York:ACM Press,2013:141-148.
[8]JAHANIRAD M,WAHAB A W A,ANUAR N B,et al.Blind source mobile device identification based on recorded call[J].Engineering Applications of Artificial Intelligence,2014(36):320-331.
[9]PANAGAKIS Y,KOTROPOULOS C.Automatic telephone handset identification by sparse representation of random spectral features[C]//MM&Sec’12,September 6-7,2012,Coventry,UK.New York:ACM Press,2012:91-96.
[10]PANAGAKIS Y,KOTROPOULOS C.Telephone handset identification by feature selection and sparserepresentations[C]//IEEE International Workshop on Information Forensics and Security,December 2-5,2012,Costa,CA,USA.New Jersey:IEEE Press,2012:73-78.
[11]KOTROPOULOS C L.Source phone identification using sketches of features[J].Biometrics IET,2014,3(2):75-83.
[12]王志锋,贺前华,李艳雄.录音设备的建模和识别算法[J].信号处理,2013(4):419-428.WANG Z F,H Q H,LI Y X.A modeling and identification algorithm of recording devices[J].Journal of Signal Processing,2013(4):419-428.
[13]邹领,贺前华,邝细超,等.基于设备噪声估计的录音设备源识别[J].吉林大学学报(工学版),2016:1-8.ZOU L,HE Q H,KUANG X C,et al.Source recording device recognition based on device noise estimation[J].Journal of Jilin University(Engineering and Technology Edition),2016:1-8.
[14]AGGARWAL R,SINGH S,ROUL A K,et al.Cellphone identification using noise estimates from recorded audio[C]//IEEE International Conference on Communications and Signal Processing(ICCSP),April 3-5,2013,Melmaruvathur,India.New Jersey:IEEE Press,2014:1218-1222.
[15]ZHAO H,MALIK H.Audio recording location identification using acoustic environment signature[J].IEEE Transactions on Information Forensics&Security,2013,8(11):1746-1759.
[16]LUKAS J,FRIDRICH J,GOLJAN M.Digital camera identification from sensor pattern noise[J].IEEE Transactions on Information Forensics&Security,2006,1(2):205-214.
[17]KINNUNEN T,RAJAN P.A practical,self-adaptive voice activity detector for speaker verification with noisy telephone and microphone data[C]//IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP),May 26-30,2013,Vancouver,BC,Canada.New Jersey:IEEE Press,2013:7229-7233.
[18]MALHI A,GAO R X.PCA-based feature selection scheme for machine defect classification[J].IEEE Transactions onInstrumentation and Measurement,2004,53(6):1517-1525.
[19]LU Y,COHEN I,ZHOU X S,et al.Feature selection using principal feature analysis[C]//The 15th ACM International Conference on Multimedia,September 27-29,2007,Baltimore,MD,USA.New York:ACM Press,2007:301-304.
[2]GUPTA S,CHO S,KUO C,et al.Current developments and future trends in audio authentication[J].IEEE Multimedia,2012,19(1):50-59.
[3]BAO Y,LIANG R.Research progress on key technologies of audio forensics[J].Journal of Data Acquisition&Processing,2016,31(2):252-259.
[4]王昌海,张建忠,徐敬东,等.基于HMM的动作识别结果可信度计算方法[J].通信学报,2016,37(5):143-151.WANG C H,ZHANG J Z,XU J D,et al.Identifying the confidence level of activity recognition via HMM[J].Journal on Communications,2016,37(5):143-151.
[5]贺前华,王志锋,RUDNICKYA I,等.基于改进PNCC特征和两步区分性训练的录音设备识别方法[J].电子学报,2014,42(1):191-198.HE Q H,WANG Z F,RUDNICKYA I,et al.A recording device identification algorithm based on improved PNCC feature and two-step discriminative training[J].Acta Electronica Sinica,2014,42(1):191-198.
[6]HANILCI C,ERTAS F,ERTAS T,et al.Recognition of brand and models of cell-phones from recorded speech signals[J].IEEE Transactions on Information Forensics&Security,2012,7(2):625-634.
[7]HANILCI C,CEMAL I,ERTAS F.Optimizing acoustic features for source cell-phone recognition using speech signals[C]//ACM Workshop on Information Hiding and Multimedia Security,June20-22,2013,Montpellier,France.New York:ACM Press,2013:141-148.
[8]JAHANIRAD M,WAHAB A W A,ANUAR N B,et al.Blind source mobile device identification based on recorded call[J].Engineering Applications of Artificial Intelligence,2014(36):320-331.
[9]PANAGAKIS Y,KOTROPOULOS C.Automatic telephone handset identification by sparse representation of random spectral features[C]//MM&Sec’12,September 6-7,2012,Coventry,UK.New York:ACM Press,2012:91-96.
[10]PANAGAKIS Y,KOTROPOULOS C.Telephone handset identification by feature selection and sparserepresentations[C]//IEEE International Workshop on Information Forensics and Security,December 2-5,2012,Costa,CA,USA.New Jersey:IEEE Press,2012:73-78.
[11]KOTROPOULOS C L.Source phone identification using sketches of features[J].Biometrics IET,2014,3(2):75-83.
[12]王志锋,贺前华,李艳雄.录音设备的建模和识别算法[J].信号处理,2013(4):419-428.WANG Z F,H Q H,LI Y X.A modeling and identification algorithm of recording devices[J].Journal of Signal Processing,2013(4):419-428.
[13]邹领,贺前华,邝细超,等.基于设备噪声估计的录音设备源识别[J].吉林大学学报(工学版),2016:1-8.ZOU L,HE Q H,KUANG X C,et al.Source recording device recognition based on device noise estimation[J].Journal of Jilin University(Engineering and Technology Edition),2016:1-8.
[14]AGGARWAL R,SINGH S,ROUL A K,et al.Cellphone identification using noise estimates from recorded audio[C]//IEEE International Conference on Communications and Signal Processing(ICCSP),April 3-5,2013,Melmaruvathur,India.New Jersey:IEEE Press,2014:1218-1222.
[15]ZHAO H,MALIK H.Audio recording location identification using acoustic environment signature[J].IEEE Transactions on Information Forensics&Security,2013,8(11):1746-1759.
[16]LUKAS J,FRIDRICH J,GOLJAN M.Digital camera identification from sensor pattern noise[J].IEEE Transactions on Information Forensics&Security,2006,1(2):205-214.
[17]KINNUNEN T,RAJAN P.A practical,self-adaptive voice activity detector for speaker verification with noisy telephone and microphone data[C]//IEEE International Conference on Acoustics,Speech and Signal Processing(ICASSP),May 26-30,2013,Vancouver,BC,Canada.New Jersey:IEEE Press,2013:7229-7233.
[18]MALHI A,GAO R X.PCA-based feature selection scheme for machine defect classification[J].IEEE Transactions onInstrumentation and Measurement,2004,53(6):1517-1525.
[19]LU Y,COHEN I,ZHOU X S,et al.Feature selection using principal feature analysis[C]//The 15th ACM International Conference on Multimedia,September 27-29,2007,Baltimore,MD,USA.New York:ACM Press,2007:301-304.