Quantization Watermarking Schemes for MPEG-4 General Audio Coding
PCM '02 Proceedings of the Third IEEE Pacific Rim Conference on Multimedia: Advances in Multimedia Information Processing
An Audio Watermarking Technique That Is Robust Against Random Cropping
Computer Music Journal
| Hi-index | 0.00 |
Nowadays the multimedia technology in distributed environments becomes realistic and the multimedia copyright protection issue becomes more and more important. Various digital watermarking techniques have been proposed in recent years as the methods to protect the copyright of multimedia data. Although, conceptually, these techniques can be easily extended for protecting digital audio data, it is challenging to apply these techniques to MPEG Audio streams because we need to design the watermarking schemes working directly in the compressed data domain. In this paper, we present two watermarking methods which will embed the watermark directly into the MPEG audio bit streams. Among the two presented schemes, the first one embeds the watermark into the Scale Factors of the MPEG audio streams, and the second one embeds the watermark into the MPEG encoded samples. The first watermarking procedure embeds the watermark to the scale factors of the encoded audio streams. In MPEG audio standard, each scale factor takes 6 bits, therefore, we have as many as 63 levels of scale factors indexed from 0 to 62 (level 63 is not used by MPEG standard). Tests show that when the scale factor is increased by 1 level (index decrease by 1) or decreased by 1 (or multiple levels), there is no significant audio distortion. We can use this observation for watermarking process. The second watermarking procedure adds the watermark (-1/1 bit sequence) to the encoded sample sequence. However, modification of these encoded samples must be performed very carefully due to the perceptual sensitivity of the human ear towards changes in audio data. Tests show that, if we change every encoded sample by either adding 1 or -1, the distortion of the resulting audio is easily detected by human ear and the watermarked audio is not acceptable. To solve this problem, we introduce a spacing parameter which reduces the number of samples to be watermarked. Users can adaptively choose different spacing parameter for different audio streams. By choosing a good spacing parameter, the distortion can be minimized. In order to design a non-invertible MPEG audio watermarking schemes, the watermark construction is also important. In our schemes, the watermark is created by first applying the standard encryption function DES to the original MPEG audio byte sequence and then simply converting all even numbers to -1''s and all odd numbers to 1''s. It can be showed that the non-invertibility property can be achieved by this watermark construction enforcement. Our experimental tests show that both methods minimize the distortion and provide good audio quality for various types of audio streams. Although we use MPEG Audio Layer II streams in our experimental tests, both of the proposed schemes can be applied to MPEG Audio Layer I streams and Layer III streams.