Watermarking is a mechanism in which owner of the audio file hides the watermark information into a audio for various applications. The identity of the owner of the audio file is hidden in the audio, which is known as watermark. In this article, a quantization‐based audio watermarking using singular value decomposition (SVD) is proposed. The original audio signal is converted into non overlapping two dimensional matrix blocks. The SVD is applied to each block. The watermark is embedded into audio signal by quantization of largest singular value of the block. The watermark is extracted blindly without using original audio signal. Experimental results show the watermark's high imperceptibility in the audio signal and good performance against Stirmark as well as traditional signal processing attacks. Compared with other audio watermarking methods, our method has higher embedding capacity and robust against various traditional signal processing attacks.

中文翻译：

---

基于奇异值分解的盲量化音频水印方案设计

水印是一种机制，其中音频文件的所有者将水印信息隐藏到各种应用程序的音频中。音频文件所有者的身份隐藏在音频中，称为水印。在本文中，提出了一种使用奇异值分解（SVD）的基于量化的音频水印。原始音频信号被转换成不重叠的二维矩阵块。SVD 应用于每个块。通过对块的最大奇异值进行量化，将水印嵌入到音频信号中。水印是在不使用原始音频信号的情况下盲目提取的。实验结果表明，该水印在音频信号中具有较高的不可感知性，对Stirmark 以及传统的信号处理攻击具有良好的性能。与其他音频水印方法相比，