Hua Guang - Digital Audio Watermarking: Fundamentals, Techniques and Challenges

(1)

(2)

(3)

Yong Xiang

Email:

Data hiding, or information hiding, is a general term that describes the problem of embedding information in a cover content [], where the former is usually associated with specific scenarios or with tamper-resisting purposes.

A brief clarification about the differences among similar terms are provided as follows. Data hiding and information hiding are considered to be interchangeable, and both terms represent the general concept of embedding a secret signal into a cover content. Watermarking, on the other hand, specifies the type of cover content as document, audio, image, video, etc., mainly multimedia content. It is also important to note that the watermarks to be embedded may or may not contain a message, that is to say, the watermarks could simply be a mark, signature or logo, instead of a message. The designer of a watermarking system is concerned with how the watermarks could survive uncontrollable processing or attacks when the watermarked content is distributed. On the contrary, in the context of steganography, the designer is more concerned with concealing the existence of the hidden massage. More details about the similarities and differences could be found in [].

The very original concept of paper marks or paper watermarks emerged hundreds of years ago []. In this book, we will provide the readers with a comprehensive review of fundamental principles, latest techniques, and open problems of digital audio watermarking. Classical audio watermarking techniques, such as echo hiding, spread spectrum (SS), quantization index modulation (QIM), and patchwork, will be introduced. Further, the development history of each type of techniques will be reviewed, including the latest works. In addition, this book covers two special topics on audio watermarking, i.e., reversible audio watermarking and audio watermarking with cryptography.

Despite the performance criteria for an audio watermarking system depend on the specific application, common performance criteria could be summarized as follows.

Generally, there exists a trade-off among the above performance criteria, and improving one would probably result in compromising another (or a few others). A typical example is the most intensively studied trade-off between imperceptibility and robustness. Conceptually, if one would like the watermarks to be more imperceptible, then he or she tends to weaken the strength of the embedded watermarks, but this will result in the watermarks being more vulnerable to attacks, and vice versa. In the context of reversible audio watermarking, robustness is not considered, while the trade-off is established between imperceptibility and capacity. Specifically, if more watermark bits are embedded, then more distortion will be introduced to the cover signal, and vice versa.

Watermarking systems were created with the original objective of copyright protection for multimedia content. Such a need has drawn more and more attentions since the era of digital technology in early 1990s. Since watermarking systems generally take effect only after copyright infringement having taken place, they are rarely used for preventing it from happening. Several companies have developed their commercial audio watermarking products to suit the need of the market. Verance []. In this application, while streaming subscribed multimedia content on a device such as a television, the subscriber could use another mobile device, e.g., a smartphone or a tablet, to receive enhanced viewing experience. In this way, immersive complementary content could be provided on the second screen of the mobile device. The communication between the second screen and the primary screen is established via audio watermarking. Specifically, audio watermarks are embedded into the audio channel of the multimedia content streamed on the primary device. Then, the second screen subscriber uses the alternative device to receive the audio signal from the primary device and extracts the imperceptible audio watermarks which trigger the service of enriched content on the second screen.