摘要
笔者选用时域音频水印算法,利用人耳隐蔽效应对原始音频文件数据进行采样分段处理,将水印图像作为指纹信息,对音频采样点数据进行修改,从而实现指纹信息的嵌入。同时为说明嵌入指纹信息后的音频文件质量良好,笔者随机选取10名测试者对嵌入水印信息的音频文件进行测听并评分,评分结果的平均值为4.2,依据MOS打分标准得到音频质量为良,基本符合要求。笔者分别计算了原始音频信号及嵌入指纹信息后音频信号的信噪比,嵌入指纹信息后音频信号信噪比由61.3713 dB降为22.6198 dB,高于20 dB,表明嵌入指纹信息后音频的透明性良好。为了能够根据用户共谋所得的音频数据找出共谋用户,笔者选取3位音频作品合法拥有者作为共谋攻击者,采取平均共谋攻击、最大值共谋攻击、最小值共谋攻击3种共谋攻击方式模拟共谋攻击过程,并分别计算指纹信息库中各个含有音频作品合法拥有者指纹信息的音频文件与平均共谋音频文件、最大值共谋音频文件和最小值共谋音频文件的相关系数,相关系数越大,表明当前被测验的音频作品的合法拥有者参与共谋攻击的可能性越高。
In this paper,the audio watermarking algorithm in time domain is selected to sample and segment the original audio file data by using the ear concealment effect.The watermark image is used as fingerprint information to modify the audio sampling point data,so as to realize the embedding of fingerprint information.At the same time,in order to show that the quality of audio files embedded with fingerprint information is good,the author randomly selected 10 testers to listen to and score the audio files embedded with watermark information.The average value of the scoring results is 4.2.According to the MOS scoring standard,the audio quality is good,which basically meets the requirements.The signal-to-noise ratio of the original audio signal and the audio signal after embedding fingerprint information is calculated.The signal-to-noise ratio of the audio signal after embedding fingerprint information is reduced from 61.3713 dB to 22.6198 dB,which is higher than 20 dB,indicating that the audio transparency is good after embedding fingerprint information.In order to find out the collusive users according to the audio data obtained by the collusion,the author selects three legitimate owners of audio works as the collusive attackers,and adopts three collusion attacks,namely,the average collusion attack,the maximum collusion attack and the mini mum collusion attack to simulate the collusion attack process,and calculates the fingerprint information of each legitimate owner of audio works in the fingerprint information database The higher the correlation coefficient between audio files and average collusion audio files,maximum collusion audio files and minimum collusion audio files,the higher the probability that the legitimate owners of the currently tested audio works participate in collusion attacks.
作者
刘天琪
靳欣欣
崔甜甜
肖科杰
张君豪
Liu Tianqi;Jin Xinxin;Cui Tiantian;Xiao Kejie;Zhang Junhao(Henan Normal University,Xinxiang Henan 453007,China)
出处
《信息与电脑》
2020年第10期38-41,共4页
Information & Computer
关键词
时域水印算法
人耳隐蔽效应
水印图像
共谋攻击
相关系数
time-domain audio watermarking algorithm
human ear concealment effect
watermark image
collusion attack
correlation coefficient
