Abstract: With the development of multimedia technology and computer network, the distribution and transmission of digital media（e.g., image, text, audio and video） become very convenient. But rampant piracy, forgery and tamper is stunting the development of digital era. How to provide available method for copyright protection and information security has become a hot research area. Digital watermarking as a copyright protection technology is developed in this condition. Digital audio watermarking is especially challenging compared with digital video and image watermarking because the human auditory system （HAS） is far more sensitive than the human visual system（HVS）. An audio watermarking algorithm based-on Modulated Complex Lapped Transform（MCLT） is proposed. Using spectrum-warping technique that shifts spectrum coefficients non-linearly, a significant binary image is embedded in audio by modulating MCLT’s angle frequency. In order to improve the security, some pretreatments for watermark are carried through before it is embedded to audio, such as repeat coding and interleaver etc. The watermarking is embedded in all audio frequency domain, and embedding capability is high. The experimental results show that the spectrum-warping watermarking algorithm is of low robustness against MP3 compression, resampling and so on. It is a fragile watermarking algorithm.