Human speech emotion recognition using CNN

Abstract

Human speech emotion recognition is known to be the procedure of identifying emotion from the natural speech. It has become necessary to understand and detect the emotions of humans through various ways to provide a better user experience for the consumers. However, individuals have a wide range of diversity in their capacity of expressing emotions. Also, it may differ depending on how those emotions are being identified. Previously, different methodologies and techniques have been used to identify human emotions. An innovative convolutional neural network (CNN) based system for recognizing human speech and emotions is presented in this paper. Using a potent GPU, a model is constructed and fed with an unprocessed speech from a specified dataset for training, classification, and testing. The overall result was 94.38% which is quite better and surpasses many other models.