Skip to Main Content
A novel algorithm for robust audio watermarking based on quantization in DCT domain is proposed. DC and AC DCT coefficient play different influence in robust and inaudibility. We consider both DC coefficient and AC coefficient to tradeoff robust and inaudibility. First, we decompose the audio into no overlapping blocks (eight data per block). Then, every block performs the DCT operation; watermark is embedded by modulating DC DCT coefficient or 4th AC DCT coefficient by turns. Every block is embedded one bit message. Experimental results show watermark is robust to common digital processing operations (add noise, smooth, compressor etc), and inaudibility is satisfaction.