A. Marcus (Ed.): Design, User Experience, and Usability, Pt II, HCII 2011, LNCS 6770, pp. 177–186, 2011. © Springer-Verlag Berlin Heidelberg 2011 Optimisation of Sound Localisation for Emergency Vehicle Sirens through a Prototype Audio System David Moore, Stephen Boslem, and Vassilis Charissis Glasgow Caledonian University, School of Engineering and Computing, Department of Computing and Creative Technologies, Cowcaddens Road, Glasgow, UK J.D.Moore@gcu.ac.uk, sbosle10@caledonian.ac.uk, v.charissis@gmail.com Abstract. This paper examines the issues associated with the localisation of emergency vehicles. A combinatory warning system is then proposed that aims to provide drivers of both civilian and emergency vehicles with a different sequence of auditory cues as well as an in-cabin warning when an emergency vehicle is in the close vicinity. For the early testing of this hybrid alert system, we used the modelling techniques currently available to the UK emergency services in order to estimate the concurrent efficiency of the siren’s auditory warnings. Keywords: Road Safety, Sound Localisation, Warning Systems, Ambisonics, Spatial Audio. 1 Introduction Emergency vehicle (EV) warning systems currently consist of three forms of alert: an audible warning, flashing lights and coloured bodywork markings. Out of the three, the most prominent system is the audible warning, which is typically based on a sweeping siren pattern. Early localisation of sirens by members of the public makes safer and swifter manoeuvring through traffic possible. However, for pedestrians and motorists alike, sirens can cause confusion, disorientation and possible danger if not reacted to within a timely manner. According to the Accident Statistics published by the British Department of Transport Road, the driver’s inability to locate the incoming EV resulted in 7 fatalities and 1,226 casualties in the UK in 2008 [1]. Today, cars have improved soundproofing and are often equipped with powerful stereos. These factors combined with the general cacophony of modern urban life are enough to mask alert sounds, making siren recognition and accurate localisation challenging. Previous work has shown that high levels of background noise can significantly affect the average driver’s ability to clearly define the position of the EV [2]. It has also been shown that the human inability to accurately locate the direction of an approaching EV is linked to the siren patterns and the limitations of the human auditory system [3]. As roads become increasingly busier it is important to design and implement systems that alert the driver of the proximity of incoming emergency vehicles and, more importantly, which direction they are approaching from. In this work we