A stethoscope's sound from heart is analyzed to diagnose a variety of diseases caused by cardiovascular disease such as heart attack, heart failure, stroke, arrhythmia, heart valve complications etc. The goal of this study is to use artificial intelligence algorithms to diagnose heart failure by classifying heart sounds. We use MFCC to convert heart sounds to images and to recognize images using the latest Google’s research called Vision Transformer(ViT). The primary process of this research is to identify and classify data related to heart sounds, which are classified into four general groups from S1 to S4. The sounds S1 and S2 are considered to be normal heart sounds. S3 and S4 are abnormal heart sounds (heart murmurs), each indicating a different type of heart disease.