Emotion recognition involves predicting or extracting feelings based on human reactions. Predicting emotions is a crucial task even without facial expressions. Human depression can lead to severe health consequences. By using a deep learning algorithm, companies can predict emotions through emotional analysis to better understand their clients’ responses to their products. Emotional analysis helps to clarify human feelings. Supervised algorithms like Support Vector Machines, Random Forests, and Naive Bayes are employed to classify emotions in the early stages. However, these algorithms often struggle with accuracy and cannot handle large volumes of data efficiently. Various companies utilize emotional analysis to enhance their understanding of clients’ responses to their products. The existing system operates with CNN and RNN models for emotion classification. While CNN does not take previous inputs into account, RNN considers the most recent input, despite its short-term memory limitations. This issue can be addressed by utilizing LSTM. The proposed system employs LSTM for text emotion recognition and CNN to analyze human facial expressions. This text classification method assesses incoming text and determines whether it falls into categories such as happy, angry, surprised, sad, or fearful. The CNN algorithm identifies facial expressions related to feelings such as sadness, happiness, or joy. Performance analysis is based on comparing the accuracy rates of both CNN and LSTM.