摘要： 提出了一种神经网络自适应扩展回声隐藏算法。利用PN序列将音频信号的单回声内核进行扩展后作为水印信号，提高了水印算法的安全性。该算法利用了神经网络的非线性映射能力确定扩展回声内核的幅值，从而避免了复杂的心理声学模型的计算过程，实现了水印嵌入的强度的自适应。仿真实验证明了该算法的有效性和可靠性。

关键词: 音频, 数字水印, 神经网络, 自适应, 心理声学模型

Abstract:

This paper proposes a neural network self-adaptive digital audio watermarking algorithm based on time-spread echo hiding algorithm. The single echo kernel of audio signal is spread by pseudo noise (PN) sequence and taken as watermark signal. Therefore, the algorithm achieves higher robustness and secrecy merits. By exploiting the abilities of neural networks and considering the characteristics of human audio system (HAS), a just noticeable differences (JDN) threshold controller is designed to ensure the strength of the embedded data adapting to the host audio itself entirely. The simulation experiment results show that the algorithm is robust to common digital audio processing methods and the quality of the audio is guaranteed.

Key words: Audio, Digital watermarking, Neural network, Self-adaptive, Human audio system(HAS)

中图分类号: 

• TP183