Detection of Gunshots from Small Arms

Abstract

This paper deals with acoustic gunshot detectionfrom small arms primarily for use in urban areas. Key part of thepaper is an examination of typical features of gunshot signals.Based on the short-term features detection algorithms areproposed in the time and frequency domains. Training andevaluation of the algorithms was performed using real gunshots,non-gunshots as well as synthetic gunshots. The non-gunshots setcontains impulsive acoustic events such as dog barking, glassbreaking, car horn, etc.