Abstract: Compressive sensing theory is efficient to sampling and reconstruction for sparse signal. In this paper, acoustic signals carried less information are sampled deploying compressive sensing and a new impact-acoustics non-destructive test method is proposed, based on wavelet decomposition and compressive sensing. The stages are building the sensing matrix using wavelet decomposition, solving the l1 minimization, getting sparsity representation in classes and computing error of representation, followed by classification to obtain the results. Simulations and experiments show that the proposed method is practical and efficient, having good performance with corrective rate above 90% under the environment SNR 3 dB.