A Study on Cross-Lingual Speech Emotion Analysis using Natural Language Processing

Emotion can be defined as a strong feeling such as happiness, anger, sadness, neutral, etc. deriving from one’s circumstances or emotion. They add color and bring many physiological changes to human life. Eccentric emotions such as anxiety, depression, and anger interfere with constructive actions and result in destructive behavior. For determining these consequences, natural language processing is widely used to recognize and analyze the speech of people used in various aspects or situations of life to understand and establish connectivity between them.Speech Emotion Recognition and Analysis help to interact with devices or people around efficiently and draw important medical conclusions, for example how a person being interviewed feels when asked a typical question. It employs a sophisticated approach to comprehend how people will react to the circumstance. With the help of vocal features along with speech, certain inferences can be made to determine the wellbeing of a person. Most of the currently used methodologies for emotion recognition are based on the acoustic features within the speech and the linguistic features are completely ignored, Also these methods make use of classifiers like SVM, Decision Tree, and Naive Bayes for the classification task. Including linguistic features along with vocal features and using a Deep neural network for classification tasks can greatly improve the performance of the emotion recognition system. This paper deals with those existing methodologies, and drawbacks of the currently used techniques and aims to better determine these characteristics to develop the emotional well-being of a person.