An acoustic abnormal detection system

Abstract

In this paper, we investigate the possibility of using emergency screams as an emergency stop button. Conventional emergency stop systems are not feasible when machines are operated by nonengineers, because they do not anything except crying in such case. We adopt wavelet transformation for pre-processing the screams, and we use a neural network for detecting abnormalities in the processed data. This paper shows several numerical experiments to demonstrate the ability of the system.<>