A Method of Increasing Watermark Information Capacity with Five Sections for Audio Watermarking Method Based on Amplitude Modification

Because digital audio data is not deteriorated by copying, its illegal reproduction can be easily made. Recently, copyright infringement becomes a social problem such as the illegal reproduction is distributed on the Internet. Hence, the audio watermarking methods, techniques to embed proprietary data into dagital audio data, have attached attention as a prevention technique against the copyright infringement. The most representative of the audio watermarking methods add their watermarks in the time domain [1]-[3], do so in the subband domain [4] and do so in the Fourier domain [5]. W.N.Lie et al. have proposed the audio watermarking method based on amplitude modification [1]. This conventional method maintains high sound quality and is highly robust to pirate attacks, including MP3 compression, low-pass filtering, time scaling, digital-to-analog/analog-todigital reacquisition and cropping. The watermark information is embedded into audio signals in the time domain. One-bit watermark information is embedded by modifying the differences of average-of-absolute-amplitude from three sections in a GOS. The principle of amplitude modification is embployed to scale amplitudes in selected sections of samples so that time-domain waveform envelope can be almost preserved. In the conventional method, the capacity of watermark information is not enough. And hence, it is desirable that the capacity of watermark information is increased. In this paper, we aim to increase the capacity of watermark information in the audio watermarking method based on amplitude modification. In the conventional method, one-bit watermark information is embedded into a GOS with three sections. And hence, the capacity of watermark informarion per section is 1/3 bits. If two-bit watermark information is embedded into a GOS with five sections, it is 2/5 bits and more than the conventional method. Hence, We propose a method of dividing a GOS into five sections in order to embed two watermark bits into a GOS. The capacity of watermark informarion per section in our proposed method is more than the capacity of watermark information per section in the conventional method. This means the proposed method can make the section longer than conventional method. Our test shows that using a long Figure 1. GOS and three sections in conventional method.