A Watermarking Method for Copyright Protection of Audio Data

Abstract

The purpose of this research is to introduce a watermarking method to protect copyrights of audio data. Watermarking is a popular topic in the field of image processing and it is widely used to protect intellectual property rights of image contents. This study will be carried out that background to determine the use of such watermarking methods for protecting copyrights of audio data. This research is all about discussing how secret information will be embedded with a transformed audio spectrum. Normally, two popular domains have been used in watermarking methods. Those could be categorized into two domains named as “time” and “transform”. Most of the related researches have confirmed that, the watermarking mechanisms in time domain are less protective compared to that of transform domain. The proposed method in this research will be based on a technique in transform domain. This research introduces a method based on Discrete Cosine Transformation (DCT). DCT converts time based spectrum of audio data to the summation of cosine waves which oscillate at different frequencies. The proposed watermarking method splits DCT spectrum into frames in a way that, the frame count is equal to the total bits of binary watermark. Binary watermark is an image that has only two possible value (0 or 1) for each pixel. Here, each frame is set to contain even count of samples. A frame is further divided into two sub frames and arrange their means in predefined order to represent watermark bit of 0 or 1. Then, whole DCT spectrum is converted back to the time based spectrum and which could be used as the watermarked audio. Embedded watermark could be extracted again by reading the mean arrangements in DCT spectrum of watermarked audio. To enable authentication, the watermark is encrypted using Data Encryption Standard (DES) in Electronic Code Book (ECB) mode. In this research, the perceptual quality of outcome is measured by Peak Signal to Noise Ratio (PSNR) and Correlation Coefficient (CC). PSNR is subjected to the imperceptibility of watermarked audio and CC are subjected to the robustness of the extracted watermark. Normally, a successful watermarking method should give higher values for PSNR and CC. Evaluations in this research describes a case study based on proposed watermarking method. This case study explains how to adjust the embedding position of watermark to optimize the perceptual quality of outcome over the proposed watermarking method. This method uses three main parameters to adjust the embedded position such as Frame Size, Frame Gap and Start Sample. Results of the evaluation shows that the proposed watermarking method could be applied for real world applications successfully.