Audio-content-based non-uniform discrete cosine transform audio reliability authentication method

Abstract

The invention provides a non-uniform discrete cosine transform (NDCT) audio reliability authentication method for generating watermarks based on audio contents. The watermark embedding process comprises the following steps of: dividing frames for original audio A, performing three-dimensional wavelet transform (3DWT) on each frame Ai, and quantifying the mean value mi of three-dimensional wavelet approximate coefficients cA3 to generate watermarks wi, wherein the quantization length is l(1)<q>; simultaneously modifying the coefficients of the cA3 so that the mean value of the coefficients is positioned at the middle point of a quantization interval, performing NDCT, and quantifying Ci<1> to embed the watermarks wi, wherein the quantization length is l(2)<q>; and finally performing inverse NDCT and inverse 3DWT on each frame, and connecting all the frames Ai containing the watermarks and the audio part A containing no watermarks so as to finish embedding the watermarks. The audio authentication method comprises the following steps of: dividing frames for audio A* to be authenticated, performing 3DWT on each frame Ai<*>, quantifying the mean value of the three-dimensional wavelet approximate coefficients cA3* by using the l(1)<q> as the quantization length to reconstruct the watermarks Wi<*>, then performing NDCT on the cA3*, quantifying the watermarks Wi<**> by using the l(2)<q> as the quantization length, calculating the normalization value of the Wi<*> and the Wi<**>, and authenticating the integrity of the audio. The l(1)<q> and the l(2)<q> need to satisfy the condition that the product of l(2)<q> x absolute value S x 2 / N is less than or equal to the l(1)<q> (N is the dimensionality of an NDCT matrix, and S is the sum of the elements of the second column). The method is suitable for authentication of data integrity in the field with high data reliability requirement.

Description

technical field [0001] The invention is an audio reliability authentication method for generating a watermark based on audio content, which is suitable for reliability and integrity authentication in digital audio transmission and storage, and can locate maliciously tampered audio locations under asynchronous attacks. technical background [0002] Today, with the rapid development of computer technology and network technology, with the continuous development of audio compression technology and computer application software technology, the transmission, copying, storage, and modification of digital audio are becoming more and more convenient and fast, resulting in piracy, infringement, Problems such as tampering have also emerged; and, due to the security factors of the transmission line, problems such as bit errors and packet loss may also occur during audio transmission, so there is an urgent need for digital audio authentication technology to identify the attribution of int...

AI Technical Summary

Problems solved by technology

The random number method, M sequence method, and chaotic sequence method of generating watermark sequence by key method are all one-bit watermark methods in essence, which cannot locate tampering. At the same time, because of the false alarm problem, they cannot meet the identification requirements of some occasions. The need for certification, etc.

Images