Watermark Synchronization: Perspectives and a New Paradigm

Synchronization is one of the most challenging elements of a watermarking system. In this paper, we survey and classify methods employed for watermark synchronization and highlight some inherent difficulties in synchronization that arise due to the nature of multimedia signals. We then propose a framework that addresses these hurdles in practical applications. Our framework utilizes feature-based embedding in concert with error correction codes capable of handling not only substitutions caused by various perturbations but also insertion and deletion events caused by erroneous feature estimates. We present a practical scheme for speech watermarking based on the framework. Experimental results show that the proposed methodology indeed enables synchronization in scenarios where a mismatch in estimated features between embedder and receiver would otherwise cause synchronization loss. We explore connections of the framework with recent theoretical analyses of watermark synchronization.