ABSTRACT

This chapter introduces sequences with an autocorrelation function of zero sidelobe such as perfect sequences and uniformly redundant arrays. It proposes a high-capacity, real-time audio watermarking technique based on the spread spectrum approach. The watermark is embedded into the modified discrete cosine transform (MDCT)-transformed permuted block with embedding locations chosen quite flexibly. Embedding watermarks in the MDCT domain is very practical because MDCT is widely used in audio coding standards. A sequence with autocorrelation of zero sidelobe is used as the direct sequence in spread spectrum communications. Digital audio signals possess special features that make it improper to calculate the similarities between two signals such as conventionally used in digital signal processing. The perceived audio similarity is commonly measured by subjective listening tests. Different types of audio clips and pseudo noise binary data are tested in the research, including music, speech, and binary data.