Novel muteness based embedding technique for robust and imperceptible audio watermarking

In recent years digital watermarking of audio signals has gained importance. Digital watermarking represents an efficient state-of-the-art technology for preventing intellectual property rights violations of audio signals by ensuring data integrity and data origin authenticity since it makes it possible to identify the source, author, creator, owner, distributor or authorized recordings. This thesis focuses on the study and implementation of an improved version of the muteness-based watermarking algorithm. The new algorithm has enhanced features when compared to the muteness based technique as repetition codes and secret keys are part of the embedding algorithm. The proposed algorithm is a time domain algorithm as the watermark is added in the muteness period of the original signal. This type of embedding minimizes the audibility of the watermark signal while maximizing robustness to attacks. The added watermark carries information pertaining to copyright protection. A better performance in terms of recovery rate is achieved by using a low rate repetition code as part of the encoder design. A detailed analysis of the watermarking algorithm is done by testing the algorithm for various single and multiple attacks and comparing them with the existing algorithms. It is found that the watermark is robust and imperceptible --Abstract, page iii