Audio Watermarking can be described in a simplified way as hiding additional information inside an audio signal. Whereas an Audio Fingerprinting process is able to recognise music or advertisements, with Audio Watermarking any number of copies of a track can be made, all of which sound the same to humans but which can be distinguished by means of the embedded information using a special software.
A typical example of a watermarking application is a music promotion service provider. The copy of the track to be sampled which every recipient receives is not a 1:1 copy but a customised copy, i.e. with its own watermark. If this title is played later on the Internet, this copy can be examined and the embedded information can be detected. The initial recipient of the piece of music is thus established.
A vital aspect of Audio Watermarking is that the audio signal itself is changed. Only in this way can a fundamental robustness be guaranteed. If, on the other hand, the information is deposited in an ID3 tag, it can be easily deleted again even without using special tools. Even data stored in the audio format’s existing fields are no longer present after a format change. The situation is different with a change of the audio signal itself. This change cannot simply be reversed and saving in a different audio format does not usually delete the watermark.
Two issues were central to the development of our watermarking process; audio quality and the robustness of the watermark. Naturally, it would be possible to develop a process that could withstand practically any attack on the watermark. However, such an approach would be at the expense of the audio quality and so would have no market potential. It was our aim, therefore, to retain the audio quality and to achieve the most robust watermark possible under those conditions.