This chapter provides a general overview of audio data hiding. The general issues are discussed first, followed by the basic techniques used to hide data in audio signals, including bit stealing, spread spectrum methods, echo methods, and quantization index modulation. This is followed by a brief description of the recent proposals presented at the Institute of Electronics, Information, and Communication Engineers of Japan (IEICE) Multimedia Information Hiding (MIH) Technical Group Meetings.
TopInformation-hiding applications are typically classified into two categories: watermarking and steganography. Figure 1 shows an outline of information-hiding technology for audio data. In Figure 1 the technology is classified by payload data and application type. As an introduction, the keywords that are usually applied in information hiding and their definitions are briefly introduced.
Figure 1. An outline of audio information-hiding technologies
The data that are embedded are referred to as the watermark or payload data. The data to be embedded are referred to as the host data or cover data. As a consequence, the watermark is embedded in the host data. The majority of the embedding and extraction algorithms require a common key. This key is essentially different from the cryptography key. It defines where to embed or the order of embedding, which is intended to conceal the payload data in the embedding algorithm by altering the amount of physical variables used in the embedding. The data generated as a result of embedding are referred to as the stego data. The stego data are expressed as quantized waveform data or encoded bit-stream data. It is generally necessary to decode the latter data to the stego waveform data using the conventional decoder of the same codec used in the encoding process. The degradation in the quality of the stego waveform must be low enough to still serve the user’s appreciation or listening purposes. The payload data is extracted by the appropriate users and is utilized for the intended purpose.
Watermark or watermarking is a term used in information hiding that is mainly applied to copy-control or digital rights management. A typical watermarking application embeds a content ID and/or a consumer ID into the host audio signal that is delivered from commercial Internet sites or conventional audio media, such as television or radio programs, compact discs, etc. If the stego signal is distributed over the Internet illegally, the crawler program in the music’s contents is able to detect the identity of the watermarked content automatically and determine where it was distributed and who purchased it.
In such an application, the extraction process of the payload data should be robust against any modifications of the stego signal as a result of the distribution of the contents. Perceptual coding for Internet distribution and digital broadcasting, bandwidth limitations, analog-to-digital and digital-to-analog conversions, and additive noise for analog broadcasting should be considered as modifications. In addition, malicious attacks such as pitch conversion and time expansion or contraction have to be considered in the context of intended illegal distribution.
In the field of image watermarking, Stirmark is a well-known tool for robustness evaluation (Petitcolas, Anderson, & Kuhn, 1998). A similar approach, Stirmark for Audio (Steinebach, et al., 2001), has also been proposed for audio watermarking. However, it is currently not popular, and it has not been applied in many studies.