This paper discusses three different aspects of audio watermarking: techniques, applications, and results of recent evaluations.

Although audio watermarking is often thought of as a new technology, many of the current techniques have been known for decades. Several of these are described.

The potential applications of audio watermarking are numerous, including the obvious copy protection and content identification functions. Some examples of applications and the implications for location of watermark embedders and detectors, are presented. The balance between payload capacity and robustness for different applications is assessed.

The EBU has recently completed tests of the robustness and subjective quality of state-of-the-art audio watermarking systems. The test methods are described, including the attacks to which the watermarks were subjected, and the subjective test method (a combination of BS.1116 and a forced choice). The results of the tests are presented.

This paper was presented at the 21st Nordic Sound Symposium Bolkesjo, Norway, October 2-5 2003.