一种可盲检测的多功能音频水印算法

为了使数字水印实现多种功能,提出了一种能在同一音频信号中同时嵌入多个水印的新方法.首先,载体音频信号被分隔成指定长度的帧,并且所有成员水印被编码成一个混合水印.然后,利用回声隐藏技术将混合水印中的二进制比特嵌入到音频信号帧中.水印提取不需要原始音频信号.为了提高水印算法的提取正确率和抵御常规信号处理的鲁棒性,功率倒谱的自相关被用于检测携带水印的音频信号帧中的回声延时以提取出混合水印,并且相应的解码算法被用于分离成员水印.计算机仿真结果表明所提出的水印算法对常规的信号处理,如Mp3压缩、重采样、重量化、低通滤波以及白噪声具有很强的鲁棒性.

Blind multiple watermarking algorithm for audio

To make digital watermarking accomplish several goals,a new method for simultaneously embedding multiple watermarks into the same audio signal is proposed.First,the original audio signal is segmented into frames of appointed lengths and all the element watermarks are encoded to achieve a mixed watermark.Then,the binary bits in the mixed watermark are embedded into the audio frames with the echo hiding technique.The watermark extraction can be performed without an original audio signal.Furthermore,in order to enhance the extraction accuracy and the robustness of the proposed algorithm against common signal manipulations,the autocorrelation of the power cepstrum is utilized to estimate the echo delays in the watermarked audio frames to extract the mixed watermark and the corresponding decoding method is applied to achieve the element watermarks.Computer simulation results indicate that the proposed scheme has great robustness against common signal manipulations of Mp3 compressing,re-sampling,re-quantizing, low-pass filtering and white noise addition.