Audio forensic marking using quantization in DWT-SVD domain

Author

De, Li ; Kim, JongWeon

Author_Institution

Dept. of Comput., Yanbian Univ., Jilin, China

fYear

2011

fDate

13-16 Feb. 2011

Firstpage

988

Lastpage

991

Abstract

In this paper, we proposed an audio forensic marking technology which can trace illegal distribution of audio contents. The forensic mark was embedded in the double transformation domain, DWT-SVD, using the quantization method for the purpose of the multi-step tracing. DWT is used for the multiple embedding and detection, and SVD is used for the embedding the forensic mark into the specific frequency domain of DWT. For enhancement of the detection performance, the block code is used. The algorithm can embed 1024bits into 32,768 sample space. When the audio signal is compressed by 128kbps mp3, the BER is 0.49% and the value of NC (normalized correlation) is 0.99.

Keywords

audio coding; block codes; discrete wavelet transforms; error statistics; quantisation (signal); singular value decomposition; BER; DWT-SVD domain; audio contents; audio forensic marking; audio signal compression; block code; discrete wavelet transform; double transformation domain; frequency domain; illegal distribution tracing; mp3; multiple detection; multiple embedding; multistep tracing; normalized correlation; quantization; singular value decomposition; Block codes; Discrete wavelet transforms; Forensics; Matrix decomposition; Robustness; Watermarking; DWT; Forensic Mark; Quantization; SVD;

fLanguage

English

Publisher

ieee

Conference_Titel

Advanced Communication Technology (ICACT), 2011 13th International Conference on

Conference_Location

Seoul

ISSN

1738-9445

Print_ISBN

978-1-4244-8830-8

Type

conf

Filename

5745974