摘要

To make digital speech watermarking achieve the goals of copyright protection and integrity verification simultaneously, a multipurpose speech watermarking algorithm that is based on the multistage vector quantization (MSVQ) of linear prediction coefficients (LPCs) is presented In the proposed scheme, a robust watermark and a fragile watermark are embedded in the first stage vector quantization (VQ) and the second stage VQ respectively with different methods. The extraction of both the robust watermark and the fragile watermark can be carried out without reference to the original speech. Simulation results indicate the effectiveness of our algorithm in terms of robustness and fragility.