Ceramic matrix composites (CMCs) show enhanced fracture toughness due to several crack deflection mechanisms, such as: matrix cracking, fiber debonding, fiber breakage and fiber pullout. The understanding of these mechanisms is crucial for the further development of CMCs. Acoustic emission (AE) monitoring can be applied to detect damage during loading of composites. However, the main challenge of this technique is to classify the measured signals to the respective mechanisms. In this work, the damage development during loading is evaluated by classifying acoustic emission (AE) signals by supervised machine learning. At first, specific mechanical tests were used to induce each damage mechanisms separately. AE signals recorded during these tests were related to their respective mechanism types and used to obtain a training dataset for a model based on the k-nearest neighbors algorithm. Model accuracy is calculated to be 88%. The model was then used to classify AE signals measured during tensile tests of different types of CMCs under quasi-static and cyclic loadings. The results show that the model can be successfully applied to identify and quantify the AE signals, providing important information about the location, time, frequency and intensity of each damage mechanism.