Reversible Data Hiding for Audio Based on Prediction Error Expansion

This paper proposes a reversible data hiding method for digital audio using prediction error expansion technique. Firstly, the prediction error of the original audio is obtained by applying an integer coefficient predictor. Secondly, a location map is set up to record the expandability of all audio samples, and then it is compressed by lossless compression coding and taken as a part of secret information. Finally, the reconstructed secret information is embedded into the audio using prediction error expansion technique. After extracting the embedded information, the original audio can be perfectly restored. Experimental results show that the proposed algorithm can achieve high embedding capacity while keeping good quality of the stego-audio.