TRANSPARENT LOSSLESS AUDIO WATERMARKING

摘要

A losslessly watermarked Pulse Code Modulated (PCM) audio signal is generated by receiving a first audio signal quantised on a first grid, mapping the samples to a third, coarser grid including reconstruction data for resolving ambiguities when the coarser grid indicates multiple first signal samples, combining this reconstruction data with watermark data and determining a second quantisation grid, and finally generating the losslessly watermarked audio signal by quantising the reconstruction/watermark signal onto the second grid. The different grids may be generated via pseudo-random sequence generators and the audio data may be multichannel. The technique allows decoding to commence partway through an audio stream by removing the dependence upon previous sample.