A Fragile Watermarking Scheme for Audio Detection and Recovery

In this paper, we propose a method for self-embedding an audio signal into itself as a means for protecting the content of audio signal. In the proposed algorithm, the host audio signal is divided into 4 segments, and the feature of every segment is embedded in the less significant bits (LSBs) of another segment that is randomly assigned based on the secret key. And the tamper detection can be performed by checking the surrounding segments of the tested segment and the segment that is carrying its feature. Experimental results show that the modified audio signal not only can be effectively detected, but also can be successfully recovered with acceptable quality.