Abstract: A new approach of combining the method of resonance demodulation and multi-resolution analysis of wavelet packet is presented to characterize the acoustic emission (AE) signals. Firstly, wavelet packet is used to decompose the acoustic emission signal into several frequency bands. Then the energy of each frequency band is calculated to draft the percentages of each band on total energy. Secondly, the above-mentioned results shall be compared with the energy spectrum of normal acoustic emission signal and the sub-bands which change more obviously shall be obtained. Based on the technology of resonance demodulation, the band which changes a lot is decomposed by Hilbert transform to get the feature of signal. According to the experimental data analysis, it shows that this method can effectively extract the characteristic of acoustic emission and it is a new way of acoustic emission feature extraction.