Nowadays many watermarking algorithms have been proposed and implemented for images and video. Compared to the human visual system, the sensitivity of the auditory system makes the audio watermarking more difficult to realize. In order to improve the invisibility of the audio watermarking,this paper presents an audio watermarking algorithm which is based on the quantization of discrete cosine transform (DCT), first the original audio is divided into two parts due to the energy, and do the discrete cosine transform on the larger part, then the Arnold scrambled binary watermark image is embedded into the mean value of the middle frequency coefficients using quantitative index modulation (QIM),And ensure that the embedding watermark on the original audio within the range of hearing range. Experimental results show that the algorithm has good invisibility, and have robust of MP3 compression and noise.