Anti-re-recording Of The Robustness Of The Audio Digital Watermarking Algorithm

With the digital and information age coming, people can quickly and easily share resources by network, storage media, etc., and freely copy and transfer resources, however, as a results, copyright protection is so difficult. At present, encryption and digital watermarking are being used to solve this problem, but in the audio field, re-recording is a new problem. When the audio being decrypted and recorded, cipher can not protect the data any more. Since the common digital watermark algorithms are poor, when faced the attack of re-recording, an audio digital watermarking algorithm based on energy radio of wavelet coefficients is proposed. This paper will be divided into several sections as following:re-recording attack analysis, algorithm design, Matlab simulation, and implement of algorithm with embedded system.1. Re-recording attack analysisAnalyze of the factor of audio re-recording:discussing time stretching, scope change, phase altering, and band-pass limiting. Describe different type transmission channel of re-recording:talking about broadband channel, narrowband channel, and acoustic coupling channel.2. Algorithm designAn audio digital watermarking algorithm based on wavelet coefficients energy radio is proposed in this paper. Cut the video into sections with certain length, divide every section into two parts, calculate the ratio of their wavelet coefficients energy, and then class the ratios into two regions statistically, where the probability of ratios falling in each region is 1/2. One region could represent the bit'0', while the other one means the bit'1'. The watermark extraction threshold could be decided by balancing the transparency and robustness.3. Matlab simulation analysisSimulate the algorithm with Matlab to analyze its robustness to the attack of signal processing, such as re-sampling, quantization, and Gaussian noise. It is proved that the algorithm is robust to the attack of digital signal processing and lightly random cutting, while its transparency is very good. 4. Implement of the algorithm with embedded systemIn order to verify the performance of algorithm, and the algorithm is realized in DSP using C language. Connect the audio input and output of the DSK to realize audio re-recording, play audio embedding watermark to audio output, read the audio data from input interface, and then extracting watermark from the data. The experiment results show that the recording delay is 27 sample points, and the bit error of watermark extraction approximately near to 0.