Abstract:

In order to distinguish different kinds of heart sound signals accurately and obtain better recognition effect, we proposed a heart sound signal recognition model based on Gauss mixture model. Firstly, wavelet transform was used to denoise original heart sound signal, [JP+1]and the interference of noise to feature extraction of heart sound signal was eliminated. Secondly, heart sound signal was extracted while model of classification and recognition of heart
sound signal was constructed by using Gauss model. Finally, the performance of the model was verified by using heart sound signal data. The results show that the average recognition rate of heart sound signal for the proposed model is more than 95%, and recognition result of heart sound signal is superior to other models.

Key words: heart sound signal, classification model, signal frame, extraction feature, liner prediction coefficient of frequency