In this paper, a new approach for human recognition using heart sounds is proposed. The new approach is based mainly on extracting features from heart sounds using wavelet packet decomposition. Different linear and non-linear filter banks at different decomposition levels are designed using wavelet packet decomposition to select the appropriate bases for extracting discriminant features. Automatic wavelet de-noising and linear discriminant analysis are adopted for pre-processing and classification stages, respectively. The proposed system is tested using an open database for heart sounds known as HSCT-11 which contains data collected from 206 subjects. Based on the achieved results, the proposed system can identify subjects with best accuracy of 91.05% and verify them with an equal error rate of 3.2%. The obtained results in this paper show that wavelet packet based features are appropriate for human recognition task using heart sounds.