The most important problem of biomedical engineering either for teaching or research activities is how to obtain data of patients with different clinical situations, because it is difficult to find unhealthy volunteers and also to get ethical approval for any experimental test on patients. Hence, this paper presents a practical generator of heart sounds and murmurs for educational and research purposes. The developed generator produces graphical representations of normal and two abnormal heart sounds, which are aortic stenosis and mitral insufficient. Three datasets of real heart sounds have been preprocessed using discrete wavelet transform, in order to reduce the noise and decompose these cardiac signals to the third approximated level based on Daubechies four-coefficient wavelet. For designation and implementation of the heart sounds generator, the ATmega 2560 microcontroller was carefully chosen to produce the output of all preprocessed datasets. The developed microcontroller-based generator showed successfully different cases of the heart sounds and murmurs, allowing to be used further for testing new automatic classification methods of cardiac valve
diseases.