Auralization of an orchestra using multichannel and multisource technique

Abstract

Previous work has shown the importance of including source directivity in computer modeling for auralizations. A newer method to capture source directivity in auralizations is the multichannel technique, which uses multichannel anechoic recordings. In this study, five-channel anechoic recordings were obtained for every orchestral part of two symphonies at the Technical University of Denmark. Five-channel auralizations were then created for each instrument, located at its typical position on-stage in a concert hall, by convolving five impulse responses from sources that each represent approximately one-fifth of a sphere with the corresponding anechoic recording channel and mixing these together. The multichannel auralizations from each instrument were subsequently combined to produce a full multichannel auralization of the entire orchestra. The results from listening tests will be presented, illustrating the perceived changes in realism, source width and depth, and source separation between the output of this full orchestral mix of multichannel auralizations and other auralization methods, including the following: (i) a mix of single-channel auralizations from each instrument, and auralizations made with a single channel orchestral anechoic recording using (ii) a surface source and (iii) a single omni-directional source. [Work supported by the National Science Foundation.]