Today watermarking of still and moving images is a wellknown technology enabling proof of authorship for the watermarked data. Similarly, there is a large interest in watermarking of audio data. Since the human auditory system is very sensitive to signal modifications, fairly elaborated schemes are required in order to perfectly/imperceptibly hide embedded data.
This paper describes how to transmit continuous data using uncompressed audio as cover data. Rather than just embedding a signature into the audio data with a watermark, a generally usable channel is provided in which any data - cryptographic or not - can be transmitted. The paper includes a description of the transmission system, quantitative evaluations of achieved bit error rates and perceptibility tests using wellknown audio quality measurements such as PAQM (Perceptual Audio Quality Measurement), NMR (Noise-to-Mask Ratio) and the ITU TG 10/4 perceptual measurement system.