Abstract: A novel watermarking scheme to ensure the credibility of audio is proposed in this paper. Psychoacoustic model controls the wavelet packet decomposition as well as the watermark insertion in the wavelet packet domain, which enhances both the transparency and the flexibility of the algorithm. Unlike previously proposed algorithms, the presented method embeds several bits of watermark instead of just one in a single coefficient. By introducing the detecting function, the new scheme can not only authenticate the integrity of the audio signal but also locate the tampered＇region and decide to what extent the audio being tampered. Experimental results show that the proposed scheme has the desired property and can be used for applications including authentication for forensic identification and news broadcasting.