Locating multiple incoherent sound sources in three-dimensional space.

Abstract

A new technology for locating incoherent sound sources in three-dimensional space is developed. The underlying principle of this technology is a model based approach that assumes acoustic radiation from a point source in free space, together with triangulations and signal pre-processing to enhance the signal to noise ratio. A prototype developed for this technology consists of six microphones, multi-channel high-accuracy data acquisition module, and a web camera. This device allows for capturing and visualizing target sources in three ways: (1) through the camera viewing angle that shows the actual sources and their Cartesian coordinates directly, (2) through projection of the source positions on a horizontal plane and their traces as they move in space, or (3) automatically switching between the camera and top view displays, depending on whether the target falls inside the viewing angle of the camera or outside. To ensure a real-time display of the results, four microphones are used. The disadvantage of using four microphones is that the accuracy in source localization may be reduced when the input data are contaminated by interfering and background signals. Under this condition, redundancy checks using six microphones can be used to enhance the accuracy and spatial resolution of source localization.