Robust frequency domain audio watermarking: A tuning analysis

In this paper, a tuning analysis of a robust audio watermarking scheme is performed. The watermarking system analysed here uses a lossy compression (MPEG 1 Layer 3) method to identify the frequencies at which the marking bits will be embedded. Several tuning parameters affect the mark embedding and the mark reconstruction processes. In previous papers, these parameters have been tuned without any systematic methodology and, thus, the capacity, imperceptibility and robustness results of the watermarking scheme could be enhanced. As a final result of the tuning analysis, some tuning guidelines are suggested for the different parameters. In addition, the conclusions of this study can be extended to other spread spectrum watermarking schemes for audio. In particular, the robustness of this watermarking scheme can be enhanced whilst improving imperceptibility and reducing capacity, which is a somewhat counterintuitive result.