Detection of Human Mental Stress Using Speech Signals

Abstract

Nowadays it is very normal for humans to experience mild or moderate mental stress in a variety of situations. A moderate level of stress is beneficial to an individual; yet, excessive stress hurts a person's mental health and is a risk factor for suicidal ideation if ignored.Long-term stress is linked to physical health concerns, according to research.With an increasing number of individuals experiencing stress, it is critical to be able to recognize it early and assist people in addressing and resolving it before significant harm is done. Conventional methods of detecting stress levels include questioning the subject and monitoring facial expression.Stress-related questions are asked throughout the interview to have a better picture of the person's condition. When people are stressed, their brows form differently, their pupils dilate, and their blinking rate may vary.These approaches have limitations in that they may overlook stress events.Research in the stress detection domain has become quite popular. There is a scope of improvement in enhancing the accuracy of the results obtained using various methods. The use of non-invasive techniques for stress detection is quite promising. This research work proposes a system to detect human mental stress using speech signals. The human speech reflects one's mental condition.The proposed research shall analyze speech signals to recognize human mental stress using machine learning techniques