Acoustic method to detect wind vane delaminating damage is a kind of feasible and relatively easy to implement, the method of acoustic signal endpoint detection is important link in acoustic resonance method to detect. In order to improve the adaptability and robustness of endpoint detection, wavelet transform and the MFCC parameters extraction are studied. Based on these research, a new parameter, DWTMFCC, is extracted. And using the SVM to audio endpoint detection. The experimental results show that, compared with the traditional wavelet and MFCC parameters, in the same noise environment, DWTMFCC has a higher rate of endpoint detection.