The heart diseases causes the death from approximately 17.7 million people in the world, and to cope with this cause of murder, it comes to predict and detect the heart anomalies as soon as possible. In order to ensure the early diagnosis, the treatment of the cardiac signals becomes as a very active research orientation. This article contributes in the field of detection of the cardiac anomalies, by proposing an algorithm allowing the detection of the S1 and S2 heart sounds from the phonocardiogram signal. The present work describes the steps followed and presents the final results of the algorithm applying on the PASCAL (Pattern Analysis, Statistical Modelling and Computational Learning) Classifying Heart Sounds Challenge. Presented results are obtained by using the homomorphic envelope with Hilbert to identify the first and second components of the heart sound. Our results are better compared to those of the three finalists of the PASCAL Classifying Heart Sounds Challenge, and exceed the results published in [3].