Mapping a Virtual Sound Space into a Real Visual Space
Résumé
The goal of this paper is to present some principles and open problems of an augmented reality system which can be used in numerous installations. This system relies on a perceptual paradox: spectators are walking into a real space such as a factory, a museum or any kind of outdoor or indoor place, seeing this space at the same time hearing through headphones a virtual sound space, homeomorph to the real one. This kind of system has several applications such as art installations, personal guided visits through a space, audio help to drivers in a reduced visibility area, audio help in the maintenance of industrial plants
.
The virtual and the real space can have any shape and physical characteristics. The only basic constraint is that the two spaces must be homeomorph, i.e. each feasible trajectory of a spectator in the real space has a corresponding continuous trajectory in the virtual sound space. Depending on the application other constraints may be added, such as the realistic or neutral or abstract nature of the virtual space acoustic. The virtual space may be "stationary" or may depend on deterministic or random past events. For example a spectator can be able to leave à "sound message" or a "sound trace" at a given place, which will be heard later by another spectator at the same place.
The technical solution proposed relies on the Spat software developed by IRCAM on LINUX. For a given spectator which head position is defined at any time by three coordinates, this software is able to compute in real time the binaural perception of the virtual sound space. In this paper we address two problems:
We presents a feasible distributed hardware and software architecture able to dynamically share the load of the signal acquisition, sound synthesis, and sound distribution according to the shape of the virtual space and the number of spectators. This architecture is under experimentation in the computer research laboratory.
We investigate the main open technical problem, which is the localization of the spectators. We discuss the precision needed according to several applications and give some possible solutions.