Audibility of characteristics in the reverberation tail of binaural room impulse responses

A binaural impulse response has a time course, where the arrival of the sound reaching the receiver directly from the source can be identified, then a number of more or less discrete reflections can be identified, and these will in the late part of the response become so numerous that they can't be separately identified. For computer generation of 3D sound it is necessary with a fairly accurate description of the direct sound and possibly some of the first reflections, but a convincing experience of the acoustical environment may also be achieved, when using an approximation for the late part of the response. In the present project it is the purpose to investigate, whether -- e. g. for a given room -- one reverberation tail can be used for the simulation, which needs not be updated if e.g. source and receiver positions vary. Since the reverberation tail as a function of time becomes more and more diffuse, this will to a great extent depend on the choice of transition time used for the division of the response into "head" and "tail". Numerous measurements of binaural room impulse responses, analysis and processing of these (see also project "Reverberation Tail Concatenation") has been carried out, and listening experiments using different combinations of "heads" and "tails" have been carried out. The results show the expected dependence of time, and that the relativ position of source and listener has an inferior role. The relation between audibility and known room-acoustical parameters is not simple, although there is some correlation to C50. The Ph.D. defense took place on December 10th, 2004, and the degree awarded January 19th, 2005. (Ph. D. scholarship for Kittiphong Meesawat by Thai government). Supported by STVF/FTP.