Audio watermarking scheme resistant to both random cropping and low-pass filtering

Common digital signal processing often bring some noise to the audio signal, and also can change high-frequency component. Therefore, the watermark embedding method through adjusting low-frequency component of audio signal can improve the robustness of watermarking scheme. Simultaneously, some signal processing operations and malicious attacks are able to change the location of watermark information. In this paper, a new robust watermarking algorithm in DCT domain is proposed. The watermark is embedded into the maximum coefficient of DCT for an audio signal, and the synchronization marker is embedded into time domain. Subjective and objective tests show the proposed algorithm is highly robust to common digital signal processing attacks such as additive noise, sampling rate change, re-quantization, MP3 compression, random cropping and low-pass filtering.