SPHEAR TMR Network Partners

6.2 Bochum

Institute of Communication Acoustics (RUB-IKA), Ruhr-Universitaet Bochum, DE

6.2.1 Experience and competence

The Institute, currently headed by Prof. Dr.-Ing. Dr. techn. h. c. Jens Blauert was established in 1974 as part of the Faculty of Electrical Engineering. Its personnel comprises 23 academic researchers (electrical engineers, acousticians, psycho-acousticians, linguists, and computer scientists), eight technical and office employees plus about 40 students who are in the process of specialising in various fields of acoustics. The Institute is a member of the Bochum-based research consortium "Telepraesenz". Activities in binaural technology started in 1975 with extensive research work in spatial hearing and its application in information technology and architectural acoustics. The Institute has developed computer models of the human binaural-hearing system with great application potential in speech technology and electroacoustics. Since about 1985, binaural-room simulation, as a major pre-requisite of auditory Virtual Reality, has become an explicit goal of research and development in the Institute. The Institute was coordinator of the successful ESPRIT- Project SCATIS (Spatially-Coordinated Audio/Tactil Interactive Scenario), a basic research project on Virtual Reality. Currently, eight full-time scientists work in the field of auditory Virtual Reality and binaural technology.

6.2.2 Role in the project

Bochum brings its leading-edge research on binaural hearing to the SPHEAR consortium. The `virtual auditory environment' technique [Hartung and Sterbing 1996 - see below] is important for a number of studies and is outlined in ?4.1.1. For auditory scene analysis, binaural processing adds the space dimension to a time-frequency representation. This makes it possible to discriminate between the desired sound source in one direction and interfering noises in other directions. Thus a recogniser can be given only that part of the binaural excitation pattern which comes from the direction of the desired source and ignores the rest (task 3.5). It has been shown [Bodden, 1996 - see below] that for two concurrent sound sources the performance of an HMM-recognizer can be significantly improved by using binaural preprocessing.

6.2.3 Research linkages

Bochum will collaborate with Patras in task 1.3, with Grenoble and Patras in task 2.3 and with Ericsson and IDIAP in task 3.5.

6.2.4 Principal research personnel, expertise and involvement

6.2.5 Recent publications