Omni-directional sources are often used in room acoustic computer simulations, as opposed to directional sources, since measured directivity data are quite limited and difficult to obtain. The purpose of this study is to investigate the objective and subjective significance of adding more complex directivity to the sources used in computer simulations and auralizations. A simple hall was used as the modelled space in the software program ODEON. Three source positions on stage and three receiver audience positions were chosen. Impulse responses (IRs) were calculated for the nine source/receiver combinations, using (a) an omni-directional source, (b) a highly directional source beaming in a sixteenth-tant of a sphere, and (c) three realistic sources: piano, singing voice and violin. The directivity data for the three realistic sources, obtained from the Physikalisch-Technischen Bundesanstalt website, were available in octave bands from 1 kHz – 4 kHz for the piano and violin, and from 125 Hz – 4 kHz for the singing voice. The objective measures evaluated were Sound Pressure Level (SPL), Reverberation Time (T60) and Clarity Index (C80). In general, there is at least 5% difference in T60 data between the omni-directional source and the realistic directional ones. Differences in SPL and C80 are more irregular across frequency bands and appear to be more apparent for sources with higher directivity index. For select source/receiver combinations, the IRs resulting from each source directivity have been convolved with anechoic musical recordings of piano, singing and violin to produce auralizations. Subjective testing revealed a noticeable difference between the omni-directional and the sixteenth-tant sources, but not with the realistic sources.
Available at: http://works.bepress.com/lilymwang/5/