A high capacity watermarking technique for stereo audio

A novel audio watermarking technique is introduced, which is based on the processing of the difference phase spectrum of stereo audio. Psychoacoustical principles of binaural hearing are employed to compute a frequency-dependent threshold of the interaural phase difference, which is used to determine locations suitable for watermark insertion. Spread spectrum is used to mark watermark locations and to provide channel synchronization. Listening tests have indicated that using this method to insert watermarks in stereo audio results in a signal that is indistinguishable from the original audio. This technique is computationally efficient and it provides payload of up to 20 kb/s making it suitable for audio information enhancement. Its fragility necessitates a friendly environment, such as in authentication applications.