Technology to Detect Levels of Stress Based on Voice Information

As a monitoring service technology, Fujitsu has developed a way to process voice signals in order to detect the level of a person’s stress based on their conversation, and also developed prototypes of a terminal—center linking system using smartphones, which are increasingly becoming widespread, to evaluate the system’s operation. This technology is intended to be used to detect problems between customers and employees at an early stage so as to improve customer satisfaction and to monitor elderly people living alone and people driving vehicles in order to ensure safety, security, etc. This technology to detect the level of stress analyzes changes in the tone of a person’s voice, that is, in the pitch and level of their voice, compared with a normal situation. Such changes occur when a person is under stress and has a dry throat. As a result of examining the detection accuracy in an evaluation experiment with simulated conversations in which the subjects were put under stress, it was found that this system can detect the level of stress with an accuracy of 90% or higher. This paper describes the technology to detect the level of stress based on the pitch and intensity of a voice and the configurations of the two prototype systems developed.