Cutomer Satisfaction Classification Based on Face Emotion Recognition and Speech Emotion Recognition

Abstract

Nowadays, customer satisfaction is very important for businesses and organizations. Manual methods such as surveys and distributing questionnaires to customers are considered less fast to get feedback from customers. Today businesses and organizations are looking for a quick way to get effective and efficient feedback on customer satisfaction, to get potential customers faster. In this study, we propose a new method for detecting customer emotion, Face Emotion Recognition (FER) and Speech snippets Emotion Recognition (SER) to identify customer satisfaction using deep learning techniques. The application is built using the Deep Learning method with CNN architecture for FER and with the ResNet architecture model for SER, the application has good performance in recognizing emotions from facial expressions and emotions that arise from speech. This can be seen from the results of the test data which produces quite good accuracy values, 73% for FER (face) and 77% for SER (speech snippets) respectively. In theory, this value indicates good classification performance.