摘要

A content authentication and tamper recovery scheme for digital speech signal is proposed. In this paper, a new compression method for speech signal based on discrete cosine transform is discussed, and the compressed signals obtained are used to tamper recovery. One block-based large capacity embedding method is explored, which is used for embedding the compressed signals. For the scheme proposed, watermark is generated by frame number and compressed signal. If watermarked speech is attacked, the attacked frames can be located by frame number, and reconstructed by using the compressed signal. Theoretical analysis and experimental results demonstrate that the scheme not only improves the security of watermark system, but also can locate the attacked frames precisely and reconstruct the attacked frames.