Three audio watermarking schemes were proposed in this thesis. First of all, a novel self-synchronization based watermarking scheme, where a special region centered at the feature point of the original audio signal was exploited to embed the watermark information. The feature point we chose can survive even after most kinds of attacks, thus the self-synchronization ability can be fulfilled through using such feature point as synchronization signal. Afterwards, adaptive embed strength and secure client-side LUT-based watermarking scheme, which divides the original audio signal into several non-overlapping segments and adaptively chooses an optimal embedding strength for individual segment according to the specific properties of each segment to embed the watermark, was presented. Secure client-side watermark is one of the possible solutions have been proposed for the copyright protection in large-scale content distribution system. Self-synchronization and LUT based duplicated watermarking scheme, which take advantage of steady and efficiency features derived from self-synchronization, duplicated watermarking and LUT-based method, reduces the possibility of the embedded watermark being removed. Besides, in this scheme, a new embedding strategy namely client-side watermarking, which is totally different from the traditional watermark embedding schemes which mostly embed the watermark at the server side before the watermarked signal was transmitted to the client side, have been applied. The adoption of the client-side strategy makes full use of the advantages of broadband technique to increase the transmission efficiency since all the copies of signal being transmitted to different clients are a unique version.