In this paper, a watermarking algorithm that uses audio signal as the watermark is presented. This approach apparently avoids the deficiencies of traditional binary watermarks that can only identify the binary presence or absence of the watermark. Embedding meaningful “signature tune” into the host signal allows the identification of ownership origin. It serves as the essential security foundation, definite meaning of the author identification. The proposed watermarking algorithm is based on Discrete Cosine Transform (DCT) domain.