the classification of cardiac signals, the methods described in the literature
to classify S1 and S2, are based on temporal criteria (duration of systole and diastole) which
will no longer be valid in several pathological cases such as tachycardia
strict. A new descr iptor from the time-frequency domain is evaluated and validated for
discriminate S1 from S2. Then, a new method of generating attributes, based on the
Empirical modal decomposition (EMD) is proposed.
Non-linear descr iptors are also tested, in order to classify sounds
normal heart and pathological sounds in the presence of systolic murmurs