In many applications of human-computer interaction, emotion prediction is essential. To enhance emotion categorization, we present a hybrid deep learning model in this study that blends convolutional neural networks (CNN) with long short-term memory (LSTM) networks. The pre-processing step refines the input data using Q-based score normalization to ensure ideal feature scale and distribution. Emotional states are robustly classified when CNN is employed to extract spatial data, and LSTM captures temporal relationships. Our model's ability to identify intricate emotion patterns is demonstrated through training and evaluation on a benchmark emotion dataset. According to experimental results, our suggested CNN-LSTM model performs exceptionally well on the test dataset, attaining 100% accuracy, precision, recall, and F1-score. These exceptional results highlight the power of combining CNN and LSTM in handling emotion prediction's spatial and continuous aspects. Q-based score normalization further enhances the model's performance by ensuring a well-distributed feature space, ultimately improving classification consistency. This study underscores the potential of hybrid deep learning architectures in improving emotion recognition applications. Our findings can be applied in diverse domains such as emotional computing, mental analytics, and human-computer interaction.
S et al. (Mon,) studied this question.