Rapid progress in computer vision and machine learning has allowed for impressive achievements in fields such as object categorization, activity detection, and face recognition in recent years. Recent research and development in these areas have allowed for these victories. Despite this, identifying human emotions remains a formidable challenge. In recent days, a lot of work has gone into trying to figure out how to fix this issue. The advancement of artificial intelligence, natural language modelling systems, and related technologies has allowed for increased precision in this response to a variety of voice and speech-based methods. As a result of its generalizability, the study of emotions has the potential to contribute to many fields. One such area is working in tandem with human computers. Customers may get insight into their feelings, make more well-informed choices, and interact more naturally with robots with the help of computers. Predicting dynamic facial emotion expressions in cinema has received a lot of interest in recent years. About 10 years ago, this pattern first emerged. This study's authors propose a deep convolutional neural networks (CNNs) model for improving the accuracy and efficiency of emotion prediction using audio clips, still images of faces, and moving images. The mel-frequency Cepstrum coefficients (MFCC) are also recovered as a feature from the speech samples supplied by the speech CNN model. The given MFCC-CNN model outperformed baseline models in the end.