Somaieh Mahdavi - University of kerman, Department of Electrical Enginering
Saeid Seydnejad
Somaieh Mahdavi
Vahid Jamshidi

In this paper, an audio watermarking method which is particularly robust against synchronization and cropping attacks is presented. Some suitable regions of the signal based on its time domain characteristics are selected to embed the watermark bits. These selected regions are converted to a transform domain. Using a secret key and a clustering algorithm some random samples within each selected region are manipulated in an iterative algorithm to embed the message bit. The iterative algorithm changes the magnitude of the selected samples of each region in time domain based on the location of the sample in the transform domain, its cluster index, and the corresponding bit of the watermark. Using temporal masking properties of HAS (Human Auditory System), the imperceptibility of embedded watermark is guaranteed. Experimental results show strong robustness of this algorithm against common audio watermarking attacks

Audio watermarking, Clustering algorithm