Because the health care field generates a large amount of data, we must employ modern ways to handle this data in order to give effective outcomes and make successful decisions based on data. Heart diseases are the major cause of mortality worldwide, accounting for 1/3th of all fatalities. Cardiovascular disease detection can be accomplished by the detection of disturbance in cardiac signals, one of which is known as phonocardiography. The aim of this project is for using machine learning to categorize cardiac illness based on electrocardiogram (ECG) and phonocardiogram (PCG) readings. The investigation began with signal preprocessing, which included cutting and normalizing the signal, and was accompanied by a continuous wavelet transformation utilizing a mother wavelet analytic morlet. The results of the decomposition are shown using a scalogram, and the outcomes are predicted using the Hidden semi morkov model (HSMM). In the first phase, we submit the dataset file and choose an algorithm to run on the chosen dataset. The accuracy of each selected method is then predicted, along with a graph, and a modal is built for the one with the max frequency by training the dataset to it. In the following step, input for each cardiac parameter is provided, and the sick stage of the heart is predicted based on the modal created. We then take measures based on the patient's condition. When compared to current approaches, the suggested HSMM has 0.952 sensitivity, 0.92 specificity, 0.94 F-score, 0.91 ACC, and 0.96 AUC.