Emotion Prediction from Text and Multilingual Voice Inputs

Emotion prediction based on the text and multilingual voice inputs is a fast-growing application area with enormous potential for improvement of the sentiment analysis, customer support and entertaining systems, diagnosis and prevention systems of emotional conditions. This work focuses on the mining of multilingual voice input and text translations for improving sentiment analysis emotion prediction. Through the integration of traditional statistical methods of ML like Naive Bayes, Support Vector Machine and current trends in deep learning, the measurement on the accuracy in predicting emotions has significantly increased. Categorized as one of the simplest and efficient learning algorithms, Naive Bayes has an accuracy of 73% on the prediction of emotion. The greatly acknowledged SVM due to its stability in the high-dimensional data analysis, earns an exclusive accuracy of 77%. Nevertheless, the classical techniques are useful, deep learning techniques especially incorporating neural networks have accelerated the field by offering much higher accuracy rates concerning the accuracy of the emotion prediction which stands at 91%. This has close connection with the advantages of applying deep learning in processing text and voice data, in which it can discover the patterns hidden from classical methods. Combining all these algorithms provide a strong framework for the emotion prediction where classical machine learning algorithms are used for the initial examination of the data while intelligent algorithms of deep learning to fine-tune the obtained results. Besides, it improves accuracy by integrating the old and new approaches and also enlarges the areas of use of emotion prediction systems in different languages and conditions. Thus, such systems are more practical in real-life and it is shown that they can substantially enhance a number of fields from customer care to mental health check from the view of predictive accuracy and combining the classical and improved method.