Audio watermarking techniques using sinusoidal patterns based on pseudorandom sequences

The spread-spectrum watermarking (SSW) technology has become an important technique that is not only widely used for still image and video watermarking, but is also used for audio watermarking. However, some technical problems, such as requiring watermark shaping to reduce audible noise, have greatly limited its application. We present a novel audio watermarking scheme using SSW technology by which we can embed a watermark into an audio signal imperceptibly with robustness to a wide range of unintended and intended attacks. The watermark is represented by sinusoidal patterns consisting of sinusoids with phase-modulation by elements of a pseudorandom sequence. We have confirmed, theoretically and experimentally, that the sinusoidal patterns based on pseudorandom sequences have the same correlation properties as pseudorandom sequences. Compared with the traditional methods using pseudorandom sequences directly for watermarking, the sinusoidal patterns are more easily manipulated for watermark embedding and detection and do not require time-consuming watermark shaping. Both watermark embedding and detection are computationally inexpensive and watermark extraction is completed by blind detection. The effectiveness of the proposed method both in inaudibility and robustness were verified by the STEP 2001 test.