Abstract: The thorough review study uses deep learning and sophisticated machine learning approaches to focus on emotion recognition through voice and facial analysis. It compiles different approaches and results from several studies in the topic. This article covers the approaches taken while utilizing Convolutional Neural Networks (CNNs) to identify emotions from facial expressions. It explores applications in Human-Computer Interaction, Healthcare, Marketing, Education, and more by looking at the utilization of different datasets and features extraction methodologies. The study focuses on the possible uses and consequences of real-time emotional intelligence in various industries. In addition, it explores in great detail the approaches taken for Speech Emotion Recognition (SER) with CNNs, assessing the performance of spectrograms and Mel-spectrograms as input types. It goes into detail on the model, feature extraction, pre-processing, and datasets that were used. Keywords: Speech Emotion Recognition (SER), Convolutional Neural Networks (CNNs), Spectrograms, Mel-Spectrograms, Audio Emotion Recognition (AER) JEL Classification Number: 031