No abstract

No abstract

No abstract

No abstract

T he advent of network and computer technology has brought about growth in the use of digital media related to electronic commerce and online services. Because digital media is easily reproduced and manipulated, anyone is potentially at risk or incurring considerable financial loss. Digital watermarking can safeguard against such loss. While the most prominent application of watermarking is copyright protection, other applications, including fingerprinting, audio authentication, and copy protection, are important research areas. 1 With the development of watermarking technologies, attacks against these systems have become more sophisticated. In general, we can categorize the attacks on watermarking systems into either common signal-processing or desynchronization attacks. 2 While common signal processing-such as MP3 compression, lowpass filtering, noise addition, and equalizationreduces watermark energy, desynchronization attacks induce synchronization errors between the original and the embedded watermark patterns and therefore can mislead the watermark detector.Most of the previous audio-watermarking schemes are robust to common signal-processing attacks, but show severe problems when faced with desynchronization attacks. To solve the problems associated with these approaches (see the ''Related Work'' sidebar for more details), we propose an audio-watermarking scheme based on support-vector-machine (SVM) theory to protect against desynchronization attacks by using audio statistics characteristics 3 and a synchronization code technique. Experimental results with SVM show that our proposed scheme is inaudible, robust against common signal processing, and robust against desynchronization attacks.