For digital audio watermark embedding based on DWT-DCT,when thewaterm arked audio is subjected tocommon attackssuch asnoise interference,low-pass filtering,and resampling,i ts robustness and security are weak.This paper proposes a new blind watermarkingembedding algorithm is propose d.After the original audio signal is preprocessed by frames,thediscrete wavelet transform (DWT) is applied to each frame,and the three-order low-frequency components are selectedfor the discrete cosine trans form (DCT).The one-dimensional signal is decomposed into two segments on average,and then the singular value d ecomposition (SVD) is carried out,and thewatermark information is embedded into thesingular matrix accordingto the characteristics ofthe singular value.In the preprocessing of watermark image,we use Arnold transform asthe basis to reduce the dimension of watermark imageto facilitate the embedding ofwatermark information. The security of digital audio watermarkingis enhanced by adding encryption key.Use 5different types of music to conduct simulation experiments.Experimental results show that the bit error rate (BER) is below 0. 0896, and the normalized correlation (NC) is above 0.9708.This algorithm has goodrob ustness to common attack methods such as noise interference,low-pass filtering,resampling,requantization,cut,and compressi on.