Analysis of increased correlation in blind audio watermark detection

Audio watermarks are often made signal-dependant to keep them inaudible in the host signals. Blind watermark detectors, which do not have access to the unwatermarked signal, may seem handicapped, because an approximate watermark has to be re-derived from the watermarked signal. Referring to the exact watermark scenario as a semi-blind detector, some reduction in performance is anticipated in blind detection over semi-blind detection. An earlier experimental work revealed that the statistical performance of the blind detector is better than that of the semi-blind detector. It is found that the re-derived watermark is better correlated to the host signal and hence, it leads to better detection performance. It is confirmed that this happens only if the embedded watermark is the same as the examined watermark. This paper focuses on an analytical treatment of the problem. The conclusions from this analysis justify the experimental observation. Clues obtained from this investigation will help in better designs of audio watermarking schemes.