Robust audio watermarking algorithm based on neighborhood averaging method

In view of copyright protection of audio works, the robust audio watermarking algorithm is proposed from the angle of the time domain. With the neighborhood averaging method, two proper positive integers are chosen to compute the mean of continuous sample values by sliding one sample every time. The two mean sequences construct two mean lines which cross each other in the time domain. Then the watermark information is embedded at these cross points with the quantization index modulation (QIM) method. The experimental results show the algorithm is imperceptible, and is resistant to common audio signal processing operations such as noise adding, re-sampling, re-quantization, low-pass filtering, MP3 compression etc. Moreover, it also survives slight time-scale modification attack. Especially, it has strong robustness to random cropping and jittering attacks.