The paper applies Hilbert-Huang Transform to process the fruit fly's wing vibration sound. Firstly, it gets Hilbert-Huang spectrum and marginal spectrum of the wing vibration sound of two different strains of fruit fly in the same species. Then, it analyzes differences of the energy size and the energy distribution, and extracts features including the relative energy, time-frequency entropy and the cumulated amplitude value of marginal spectrum. Finally, recognizes two different strains of fruit fly by BP neural network and obtains a satisfying result. The experiment result verifies effectiveness and feasibility of the feature extraction, and the paper offers a new evidence for the study of discrimination in intraspecies relationship of fruit fly.