A kind of multifunctional audio digital watermark algorithm was put forward according to the copyright protection of audio problems. The audio signal is layered three by wavelet packet decomposition and robust watermark is embedded in the low frequency part, fragile watermark is embedded in the intermediate frequency part. The experimental results show that, the algorithm has no sound very good, excellent robustness and fragility. In the actual use of the process, which not only guarantees the safety of audio information effectively , but also guarantees good audio information fidelity.