The domain of Speech Emotion Recognition (SER) has experienced a tremendous revolution due to the outbreak of deep learning, which has contributed, as in many other research areas, to a significant boost in terms of model accuracy. SER refers to a branch of Human-Computer Interaction (HCI), which deals with recognizing emotional states from human speech. Although being a thriving field of research, SER still poses a number of non-trivial challenges, mainly due to the lack of shared best practices and highquality datasets that can make the developed models suitable for their application in real environments. In this paper, we implement a CNN-based model combined with a Convolutional Attention Block, and conduct a series of experiments involving a selection of four English datasets popularly used for SER applications: RAVDESS, TESS, CREMA-D, and IEMOCAP. After testing the proposed pipeline on individual datasets, achieving a mean accuracy of 83%, 100%, 68% and 63% respectively, we perform an extensive crossvalidation between common emotional classes belonging to single datasets or combinations of them, with the aim to investigate the generalization abilities of the extracted features.INDEX TERMS Speech emotion recognition, affective computing, deep learning.