A robust multiple audio watermarking algorithm based on the combination of modified DWT and LSB is proposed by this paper. It is used for embed the watermark into audio signal for different purpose and different protection target since this algorithm combines the discrete wavelet transform and the least significant bit theory. The experimental results show that the algorithm achieved not only good resistance to normal attack but also to the shear attack. The robust watermark based on this algorithm reflects an excellent robustness since even under strong cropping attack it still can be extract clearly from the host audio. At the same time, the vulnerability watermark is distorted even under tiny disturbance situations that can be used as a cue for tempering the data.