Human–computer interaction (HCI) has become an inseparable part of modern life. While text-based HCI is reaching a high level of maturity, there is still significant room for improvement in our communication with machines. This work explores extracting emotional data from human speech as a potential enhancement to Natural Language Processing (NLP) methods. A comprehensive method for the Speech Emotion Recognition (SER) system is provided, with a focus on applying the Mel-Frequency Cepstral Coefficients (MFCC) method to extract emotion-related features from audio signals. Two different machine learning models are compared in terms of accuracy and training time using various model parameters. The results indicate that among the tested models, the Support Vector Machine (SVM) with an MFCC sample rate of 17,640 and 128 MFCC features achieves the highest accuracy of 80.56%.