Abstract: In order to solve the problem of illegal transmission and use of audio content after rerecording, a robust audio watermarking algorithm was proposed. Based on discrete wavelet transform, Discrete Wavelet Transform-Coefficient Logarithmic Mean (DWT-CLM) was defined, the relationship between DWT-CLM and the approximate components of DWT was analyzed, and a watermark embedding method was proposed based on DWT-CLM. Experimental results showed that the proposed algorithm had the good inaudibility and anti-signal processing ability, especially in the signal processing to meet a certain robustness, and a certain tolerance for re-recording attacks.