Audio Watermarking Algorithm Against Speed Attack based on Amplitude Rearrangement

[Purposes] Digital audio watermarking technology refers to the technique of embedding a watermark signal with specific meaning into the original audio signal, which can be extracted when needed. This technology has important applications in copyright protection, content authentication, and secure communication. With the increasing maturity of audio editing softwares and online audio editing websites, it is becoming more and more convenient to modify audios. Audio speeding-up is a common modification, but existing digital audio watermarking algorithms need to be improved in the robustness against audio speed-up attacks, so as to resist this kind of modification well. To address this problem, an audio watermarking algorithm based on waveform similarity overlap and amplitude rearrangement is proposed. [Methods] Based on the principle of speed-up, the proposed algorithm realized dividing of the audio region by waveform similarity overlap method, and determining of the invariant region that exists before and after speed-up as the embedding area to ensure the stability of the watermark in speed-up attacks. After discrete cosine transform being performed on the embedding area, the mid-frequency area that is not easily perceptible and has good anti-attack properties was selected, and the amplitude rearrangement method was used to embed the watermark, so that the total energy of the area remained unchanged before and after embedding, ensuring the imperceptibility while improving the robustness of the watermarking algorithm against speed-up attacks. [Results] Experimental results show that the algorithm can improve the robustness against speed-up attacks while maintaining the imperceptibility of audio watermarking, with an average bit error rate less than 0.02.