Abstract

We are all exposed to stress in our day-to-day lives due to our busy schedules, to achieve work-life balance, perform daily household chores, manage kids, at the workplace, etc. Sometimes, we tend to neglect the development of negative emotions in us which might cause damage to our physical as well as mental wellbeing. Facial expressions are a nonverbal way of communication, unless and until it is forcefully suppressed, in usual cases, human faces exhibit their emotion through expressions and the use of computer vision-enabled applications would help every individual to identify stress by analysing one’s facial expression.

In addition to facial emotion recognition, if speech emotion and text emotion can also be recognized, it would be more accurate and beneficial to individuals. The motivation behind choosing this research study is the increasing stressful events in a common man’s life which could be identified and managed to an extent by developing an application with emotion recognition capabilities.

The current research study helped to develop a CNN model with an accuracy of 98.4% and 73.67% for the Speech Emotion Recognition system and 71.78% for the Facial Emotion Recognition System. Emotion from the text was extracted using the python package, Text2emotion.

A Multi-Modal Emotional Recognition Model