Wednesday, April 22, 15:00 - 17:30, Location: Show and Tell Area A
Banu Gunel, Huseyin Hacihabiboglu, Ahmet M. Kondoz
Acoustic sound source separation has various applications ranging from hearing-aid devices to sound recording, teleconferencing to surveillance. In general, these applications require that the size of the microphone array and the number of sensors it contains are small for easy interfacing, data capturing and integrating with other systems. The processing should also allow real-time operation and the separation performance should not depend on the source direction with respect to the array. This demonstration will introduce an acoustic source separation system that meets these requirements. The microphone array used in this solution is compact, made up of four sensors, and symmetric in three dimensions. The array is used for calculating the intensity vector directions whose statistical properties are exploited for spatial filtering in order to achieve real-time acoustic source separation. The system also includes a multichannel sound card for interfacing to the associated software running on an off-the-shelf laptop. The software processes the recordings and enables the user to select the sources for separation through a graphical user interface.