Audio Information Hiding Based on Spatial Masking

Herein an information hiding technique for audio signals is proposed based on the spatial masking phenomenon observed in the human auditory system. This method employs an audio encoding technique called Ambisonics, which has recently received much attention because unlike conventional audio formats that cannot retain precise spatial information, Ambisonics is capable of reproducing sound space. For secrecy, information is hidden as an additive phantom sound source, which should be located near the real one to exploit spatial masking. Thus, only individuals who know the position of the phantom sound source and the key sequence can extract the hidden information. Additionally, computer simulations of common signal processes confirmed that the proposed method is robust against the all tested attacks except for the pitch scaling attack.