Abstract: An adaptive robust audio blind watermarking algorithm based dither quantization and signal to noise ratio is proposed.Based on analyzing open secure question of classical quantization means,dithering quantization is proposed to embed and extract watermarking.Dithering vector is generated by mapping chaos sequences which is controlled by keys,so system security is improved.To aiming at the problem of existing watermarking algorithms to ascertain the appropriate strength by repeating experiment or experiential data,the proposed algorithm calculates adaptively watermarking strength by SNR formula based on audio content.Experimental results prove the proposed algorithm is inaudible and robust for common signal process manipulations.