Abstract
Digital watermarking is regarded as an effective method for copyright protection of digital multimedia files. This paper proposed a new algorithm that hides the binary information in audio signal by modifying magnitudes of harmonic components. The harmonic components are significant in the auditory perception and are supposed to be more robust than the masked area, which is utilized in conventional audio watermarking techniques. Relative algorithms are also developed to make it possible to process a complete audio data automatically. The experiment results confirm that the watermark has a recovery accuracy above 90% after the watermarked data is attacked by most commonly used signal processing, such as digital-analog conversion, low-pass filtering, and lossy compression.