An orchestra can be simulated in room acoustics computer modelling using a variety of methods, ranging from a single omni-directional source to individual sources of all instruments. This study utilizes the method of individual sources for each instrument, but with reduced source representation for the string sections. The anechoic recordings used in this investigation are five-channel recordings, which capture the source directivity of the individual instruments. For each string section, the individual anechoic recordings were phase shifted several times, up to 23 ms, and combined to create a single recording for use in the simulations. An orchestra was simulated in three different configurations – American (first and second violins adjacent), European (first and second separated) and a completely random arrangement. For each configuration, auralizations were created using a single channel and five-channel representation for each instrument or section and for both a Brahms and Mozart symphony. Listening tests were conducted to determine if subjects could detect differences in auralizations created using the three different orchestra configurations. Preliminary results from this pilot study show subjects can detect differences between some of the configurations, particularly the American versus Random, and European versus Random, with more of an effect with Brahms than with Mozart.
Available at: http://works.bepress.com/lilymwang/25/