A Robust Audio Watermarking in Cepstrum Domain Composed of Sample's Relation Dependent Embedding and Computationally Simple Extraction Phase

ABSTRACT

Watermark bits embedded in audio signals considering the sample’s relative state in a frame may strengthen the attack-invariant features of audio watermarking algorithm. In this work, we propose to embed watermarks in an audio signal considering the relation between the mean values of consecutive groups of samples which shows robustness by overcoming common watermarking challenges. Here, we divide the host audio signal into equal-sized non-overlapping frames which in turn is split into four equalsized non-overlapping sub-frames. After, transforming these sub-frames in cepstrum domain we finally use the relation between the differences of first two sub-frames and last two sub-frames to embed watermarks. Depending on the watermark bit (either 0 or 1) to be embed, our embedding technique either interchange or update the differences between these groups of samples by distorting the sample values in sub-frames selectively. Thus, watermarks are embedded by making a little or no distortion of the sub-frames which helps our scheme to be imperceptible in nature. Moreover, use of such embedding technique lead our watermarking scheme to a computationally less complex extraction method. Simulation results also justify our claim of the proposed scheme to be both robust and imperceptible.

KEYWORDS

Audio watermarking, copyright protection, cepstrum domain, differences of mean, frame, group of samples, mean, relative states, sub-frames, transform domain