Analysis on audio watermarking algorithm based on SVD

Many audio watermarking schemes based on singular value decomposition(SVD) have been proposed in these years. Original data is usually processed as follows in existing literature. First, original data is rearranged into 2-dimensional matrix, then, SVD is applied on it, and watermark information is embedded by modifying singular values. Analysis is performed on the embedding method described in an existing literature, and the law on modification of original data during embedding procedure is deduced. The computational complexity in embedding and extraction procedures is also discussed. Our work is helpful to find the deficiency of watermarking algorithm based on SVD, and propose better watermarking scheme which may maintain high audio quality.