Heart sounds are created during the cardiac cycle, blood flows into the heart chambers as the cardiac valves open and close. The blood flow creates auditory sounds; the more turbulent the blood flow, the more vibrations are produced. In healthy adults, there are two normal heart sounds that occur in sequence with each heartbeat. These are the first heart sound (S1) and second heart sound (S2), produced by the closing of the atrioventricular valves and semilunar valves, respectively. The cardiovascular diseases (CVDs) are the number one cause of death globally: more people die annually from CVDs than from any other cause. An estimated 7.2 million deaths were due to coronary heart diseases. This study is developed to produce a deep learning model to detect signs of heart diseases through classifying heart sounds. The anticipated method would serve as an initial screening of cardiac diseases which can help in detecting signs of heart diseases. The results of the screening can be used by pathologies of both a hospital environment by a doctor (using a digital stethoscope) and at home by the patient (using a mobile device). Our classification Heart Sounds model uses Deep Learning (DL) techniques using spectrogram by converting audio data to images that utilize the advantages of Mel-Frequency Cepstrums (MFC) to extract perceptual features Mel Frequency Cepstral Coefficient (MFCC). The results improved results and a significant reduction amount of training and testing times using evaluation metric including Accuracy, Loss, Precision, Recall, and F1-Score against including VGG16, ResNet, MobileNet, Inception V3 and Xception. The proposed model attained 100% F1-score accuracy and 100% testing accuracy.