In this paper, a particle swarm optimization (PSO) based audio watermarking scheme is proposed. The watermark is embedded by modifying singular values of the host audio signal with the spread spectrum method. The designed optimization technique employs PSO to search for optimal watermark strength by keeping adapting the fitness values so that the conflicting requirements of robustness and imperceptibility of the watermark is optimized. Moreover, the proposed scheme has the flexibility to add or drop various watermarking attacks for optimization. The experimental results show that the proposed algorithm yields a watermark that is visually recognizable and very robust to typical signal manipulations. The experimental results are also compared with the results of recent work using genetic algorithm (GA) optimization technique. Comparison with GA method indicates that the proposed algorithm is more robust against most common attacks.