“…Introduction: Accurate and reliable recognition of the fundamental heart sounds (FHSs) in a cardiac cycle of the phonocardiogram (PCG) signal plays a vital role in valvular split, cardiac stress, pulmonary artery pressure analysis, left ventricular pressure rise (LV dP/dt) measure, non-invasive blood pressure estimation and human identification [1][2][3][4][5]. In the past years, many S1/S2 sound recognition methods were presented based on variety of preprocessing and feature extraction techniques and classifiers such as logistic regression-hidden semi-Markov model (HSMM) and wavelet transform (WT) [4], mel-frequency cepstral coefficients (MFCCs) and deep neural networks (DNNs) [6], multifractal decomposition [7], deep convolutional neural network (CNN) and support vector machine (SVM) [8], ensemble empirical mode decomposition (EEMD) and kurtosis features [9], total variation and Shannon energy (SE) [10], S-transform [11], empirical mode decomposition (EMD) WT algorithm with SE [12], Hilbert transform and adaptive thresholding [13], expert frequency-energy based metric [14], matching pursuit [15], duration-dependent hidden Markov model (HMM) [16], sound energy [17], high-frequency signatures (HFSs) [18], homomorphic envelogram and self-organising probabilistic model [19], adaptive sub-level tracking and Shannon-energy tracking [20], probabilistic models [21], time-delay neural network (NN) [22], NN and conventional classifiers [23], and spectral tracking [24].…”