Optimized SS-based Audio Watermarking System Using Partial Embedding Method

In recent years, digital audio watermarking has shown increasing academic and commercial value. The embedding method in frequency domain with spread spectrum technology performed well in infomation-hiding communication. But when applied in copyright protection, it doesn't have a good performance. we propose an optimized embedding method, in which watermark information are embedded only on the imaginary part of FFT points (frequency lines) instead of on both imaginary and real part. The optimization makes the watermark hardly perceived by people, meanwhile, it gives the watermark great robustness against attacks especially from perceptual audio compression. The synchronization codes are also proposed in this paper as an approach to adapt to desynchronization attacks like possible audio cropping.