AUDIO WATERMARKING FRAMEWORK USING MULTI-OBJECTIVE PARTICLE SWARM OPTIMIZATION

Aiming at the multi-objective essence of optimal audio watermarking problem, we propose a novel audio watermarking framework in this paper, which can optimally balance all conflicting objectives of the problem, fidelity and robustness against different attacks. In the proposed framework, a multi-objective particle swarm optimization technique based on fitness sharing is applied to search optimal watermarking parameters and Pareto-optimal solutions are used to express the optimal parameters found. In addition, the proposed framework has the following advantages: (i) it can avoid the difficulty of determining optimal weighted factors in the existing single-objective watermarking schemes; (ii) Pareto-optimal solutions can offer the flexibility to select optimal parameters for satisfying different application demands.