Perinatal Asphyxia is causing the death of about 1.2 million newborn infants every year. It is one of top three causes of infant mortality in developing countries. The current way of determining the occurrence of perinatal asphyxia is by the analysis of a blood sample, something requiring medical settings and competent staff, things which often lack in rural areas of those countries. That lack usually leads into late detection of the illness, resulting into brain damages or even death of the concerned infants. The initial step of the project was to develop a prototype for the perinatal asphyxia diagnosis and reproduce the state-of-the-art results found in the literature using machine learning on infant cry samples. The next step was to try another kind of features than what was previously used and the final step was to compare the results. We designed a support vector machine (SVM)-based pattern recognition system that models patterns in the cries of known asphyxiating and non-asphyxiating infants, using Mel-frequency cepstrum coefficients (MFCC) and scattering coefficients. We had only 6 samples of distinct asphyxiating infants and thus a system has been designed to take this issue into account. For the time being, we cannot conclude anything as the database needs to be expanded. Taking into account the segmentation gives us an average accuracy of 87.5% for both the MFCCs and the scattering coefficients.