A novel approach is proposed for robust audio water- marking in wavelet domain. It emphasizes on enhancing security by dynamically modifying embedding strategy. The modification is based on real-time changes of the water- mark information and host audio. Without using any secret key, informative watermark are embedded into audio with different strength. The distribution of watermarked audio in wavelet domain is relatively similar to clean audio, com- pared to existing watermarking schemes in wavelet domain. The analysis and experimental results are given to demon- strate that the proposed watermarking scheme is effective against common signal processing manipulations and at- tacks, such as Gaussian noise corruption, resampling, re- quantization, MP3 compression and D/A-A/D conversion.