Human Speech and Text Emotion Analysis : A Survey on Emotion Analysis using Deep Neural Networks

• Published in: 2021 5th International Conference on Computing Methodologies and Communication (ICCMC) pp. 1176 - 1183
• Main Authors: Singh, Sweta, Badoni, Abhishek, Deeba, K.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 08.04.2021
• Subjects: Computational modeling; Computer architecture; Convolution; Emotion Analysis; Focusing; Machine Learning; Neural Networks; Recurrent neural networks; Social networking (online); Speech; Support vector machines; Text
• DOI: 10.1109/ICCMC51019.2021.9418474

Human Emotion Analysis is an extremely wide area of study which can have several implementations for various applications. To work on any kind of project related to Emotion detection it is imperative to collect information on the available models, datasets and performance statistics related to them. This review analysis aims to consolidate a set of unique as well as state-of-the-art approaches that have been proposed and/or carried out by various individuals in this field. The proposed research work has consolidated studies across different datasets majorly focusing on the standard The Interactive Emotional Dyadic Motion Capture (IEMOCAP) dataset and Social media datasets. Our survey spans multiple modes of data such as text, speech, speech transcriptions [6] and motion capture [10] to provide a more in-depth analysis of the accuracies and improvement achieved across mentioned modes of data. Models based on Convolution Neural Networks (CNN) [6][8] and Recurrent Neural Networks(RNN) [10] have been reviewed upon among others reflecting the efficiency of each model.