Classification of Human Emotions Based on Voice Patterns to Improve Accuracy Using Convolutional Neural Network Over Random Forest

The purpose of this research was to develop a model that can classify human emotions based on voice patterns using a CNN and compare its performance with the Random Forest algorithm. To achieve this, we used the CNN algorithm with a dataset split into 70% for training and 30% for testing. In total, 20 samples were used, and the study was conducted with an alpha value of 0.05, a beta value of 0.2, a 95% confidence level, a G power of 80%, and an enrollment ratio of 1. After running the classification tests with both algorithms, we found that the Novel Convolutional Neural Network achieved a higher accuracy of 94.95% compared to the Random Forest's accuracy of 91.50%. This significant difference in accuracy is supported by a statistical analysis showing a p-value of less than 0.01, indicating a strong significance.