Estimating Crew Alertness From Speech Available to Purchase

With the latest initiative of the government to develop a high speed passenger rail system in the United States the first and most important strategic transportation goal is to “Ensure safe and efficient transportation choices. A key element of safe railroad operation is to address the issue of fatigue among railroad operating employees and how to fight it. In this paper, we are presenting a novel approach to estimating fatigue levels of train conductors by analyzing the speech signal in the communication between the conductor and dispatch. We extract vocal indicators of fatigue from the speech signal and use Fuzzy Logic to generate an estimate of the mental state of the train conductor. Previous research has shown that sleeping disorders, reduced hours of rest and disrupted circadian rhythms lead to significantly increased fatigue levels which manifest themselves in alterations of speech patterns as compared to alert states of mind. To make a decision about the level of fatigue, we are proposing a Fuzzy Logic algorithm which combines inputs such as word production rate and speech intensity to generate a Fatigue Quotient at any moment in time when speech is present. The computation of the Fatigue Quotient relies on a rule base which draws from existing knowledge about fatigue indicators and their relation to the level of fatigue of the subject. For this project, the rule base and the membership functions associated with it were derived from real time testing and the subsequent tuning of parameters to refine the detection of changes in patterns. It was successfully shown that Fuzzy Logic can be implemented to estimate alertness levels from speech metrics in real-time and that the membership functions for this purpose can be found empirically through iterative testing. Furthermore, this study has proven that the framework to run such an analysis continuously as a monitoring function in locomotive cabins is feasible and can be realized with relatively inexpensive hardware.