Binaural modeling and auditory scene analysis

During the past years auditory scene analysis has become a focus of research on the human auditory system. The abilities of humans to analyze their auditory environment are so striking that they are attracting researchers from different areas. Traditionally, research on this topic was initiated and mainly performed by psychologists. But, besides the main interest to understand the underlying mechanisms, engineers are getting more and more involved. The reason for this is that possibilities to simulate auditory scene analysis offer new solutions to a variety of technical problems. Due to that, the model presented in this article, which gives a review of the work performed at the Institute of Acoustics of Bochum University in Germany, is mainly motivated by this problem-oriented thinking. In contrast to the majority of activities in the field of auditory scene analysis our approach is based on binaural peripheral processing. The intention is to understand and simulate binaural hearing. As a result of this work several models of the binaural auditory system have been developed, ranging from closely physiologically oriented models to rather signal-processing-motivated algorithms. The focus of work is on applications of these models to technical problems. These applications can be grouped into two topics: sound source localization and simulation of the cocktail-party-effect. This article describes a current model, shows its applications, and discusses its relation to auditory scene analysis.