WATERMARK INSERTION AND EXTRACTION

摘要

A high-quality signal, for example a unity-bit coded (DSD) audio signal having a 2.822 MHz bit rate (fs1) is converted to a lower sample rate (fs3) PCM signal by means of a sample rate converter (102). The watermark is embedded at the lower sample rate by a conventional watermark embedder (101) which is capable of handling signals at this lower rate. The watermark (WM) is not otherwise available. It is subsequently retrieved by subtracting (103) the unwatermarked signal from the watermarked signal, and up-sampled (104) to an intermediate sampling rate (fs2). The DSD signal is converted (110) to a PCM signal at said intermediate sample rate. The retrieved watermark is then added (107) to the PCM signal and the watermarked PCM signal is back-converted (120) to a unity-bit coded DSD signal. In a preferred embodiment, the arrangement includes a compensation circuit (105) which compensates the information signal X' for any (e.g. non-linear) operations performed by the embedder (101) so as to minimize the estimated watermark (WM'). The compensation circuit re-introduces said operations in the DSD domain by controlling parameters such as scaling (106, 109) and time shifting (108).