An audio watermarking scheme based on VQ codebook pairing

An audio watermarking scheme based on VQ codeword pairing is proposed. The original audio signal is segmented into frames and the watermarking information is embedded into several frames. The embedding operation is conducted by modulating the middle-frequency coefficients after MDCT transform according to the watermarking sequence and the genus bits of subcodebooks. The subcodebooks are formed by codeword pairing. The extraction of watermarking needs no original audio signal and only demands to determine which subcodebook the matched codeword of middle-frequency coefficients belong to. Experimental results show that the proposed scheme has good performance on imperceptibility and has robustness to the normal attacks of lossy compression, addnoise and so on.