Characterization of $S_1$ and $S_2$ Heart Sounds Using Stacked Autoencoder and Convolutional Neural Network