Huang Analysis of DA and AD Conversions in QuantizationBased Audio Watermarking.pdf

International Journal of Network Security, Vol.3, No.3, PP.230–238, Nov. 2006 230 Analysis of D/A and A/D Conversions in Quantization-based Audio Watermarking Shijun Xiang12 and Jiwu Huang12 (Corresponding author: Jiwu Huang) School of Information Science and Technology, Sun Yat-Sen University1 Guangzhou 510275, P. R. China. (Email: isshjw@) The key laboratory of Information Security Technology, Sun Yat-Sen University2 Guangzhou 510275, P. R. China. (Received July 16, 2005; revised and accepted Aug. 12, Sept. 5, Oct. 4, 2005) Abstract corruption, MP3 compression, re-sampling, re-quantizing, DA/AD. Due to its advantages, quantization-based embedding has Most of the recent audio watermarking algorithms been introduced into audio watermarking to improve ro- can be grouped into two categories: additive and bustness performance. Existing audio watermarking al- quantization-based watermarking schemes. The additive gorithms often focus on a given attack. However, in some scheme [6] embeds the watermark in time domain [1] transmission environments, digital audio files may suffer or frequency domain [5], while in the quantization-based from the different attacks. For example, effects of D/A scheme [3, 8, 11, 12, 22, 24], the original signals are quan- and A/D conversions (denoted as DA/AD in this paper) tized by different quantizers varied with watermark infor- on audio watermarking may be modeled as modification mation to embed watermarks r