Relations with ability for driving and the utterance sound

Abstract

1.PurposeIn recent years, the traffic accidents by the advanced age driver suspected to have cognitive impairment becomes a social problem. We developed a new method that predicts ability for driving from the voice that the person uttered. This method using the voice does not need specialized equipment, and there is the advantage that we can perform routinely, remotely and easily. 2.MethodAbility for driving by the driving simulator inspected 45 elderly people as a subject. In addition, we recorded their voice by the fixed form sentence reading aloud. We carried out machine learning as learning data with large number of voice features and driving test values and derived a classification model and a regression model.3.ResultsThe classification of the two level and the prediction by the regression model showed good results together.