A Robust Blind Audio Watermarking Scheme Based on DCT-DWT-SVD

Abstract

In this paper, Watermarking hybrid sound algorithm is presented to protect the copyright of audio files, which, in addition to the clarity and consistency of the audio signal, has increased the strength and strength. To this end, a hybrid algorithm for voice signal cryptography is presented in the three domain parser transforms, discrete cosine transform, and discrete wavelet transforms. So, after discrete cosine transformation (DCT) on the host signal, by selecting the sub-band of low frequency, which contains the highest signal energy, two discrete wavelet transform (DWT) with a random wavelet filter on the low-frequency coefficients of conversion A discrete cosine applies, after selecting the approximation coefficients, the resulting one-dimensional matrix is converted to a two-dimensional matrix, and finally the resulting matrix is applied to a single value decomposition (SVD), which results in the formation of a The diameter matrix is that the watermark bits are embedded in the first layer of the dipole matrix, so that the two bits with value S (1,1), S (2,2) of the matrix The diameter S is chosen, first compares the first and second intersections of the diameter matrix S, which is multiplied by the coefficient θ multiplied by the obtained two bits and is used as a fixed value in the embedding formula and The title of the new watermark is embedded in S (1,1). The results of the implementation show that the proposed algorithm succeeded not only in achieving transparency and resistance to general audio processing attacks, such as Gaussian white noise, quantization rates, decreasing and increasing the rate of sampling, compression and low pass filtering. But has achieved better results than other similar algorithms.

Keywords