IoT-Based Encrypted Speech Emotion Recognition in Cloud

In this paper, we propose a secure cloud-based automatic speech emotion recognition system in which the speech signals are capturing using Internet of Things (IoT) devices from a user. Before dealing with cloud, a hybrid and hyper chaotic encrypted server used to ensure the user privacy. In the cloud, a previously trained Convolutional Neural Network (CNN) is used for the selection of the learning features from the speech signals. A support vector machine (SVM) included with deep learned speech features is introduced for the classification of the input in a corresponding speech emotion class. Two publicly available databases namely the eNTERFACE05 database and TESS database used to perform the evaluation. The subjective evaluation of the proposed design gives the validation of the encrypted and decrypted speech signals. Also the speech emotion recognition performance of the proposed design outperformed the state-of-the-art research works in terms of accuracy.