In this work, we present a method to extract features from heart sound signals in order to enhance segmentation performance. The approach is data-driven, since the way features are extracted from the recorded signals is ...