Cardiac Sound Classification Using Mel-Frequency Cepstral Coefficients (MFCC) and Artificial Neural Network (ANN)

Auscultation of heart sounds is usually used as an important way to identify symptoms of heart disease. In this case the experts need to concentrate on diagnosing heart sound abnormalities in humans. Finding various characteristics to classify heart sounds according to the group is a very important part. This research was made to improve the results based on previous research which still had an accuracy rate of 92%. Heart collection obtained from the Michigan Sound Heart Database. Data used only on the apex. The Mel-Frequency Cepstral Coefficients (MFCC) method is used to extract heart sound features, while for classification from the results of heart sound extraction using the Artificial Neural Network (ANN) method. Based on the results of this study, the accuracy reaches 100%.