Objective and subjective evaluation of reflecting and diffusing surfaces in auditoria
The performance of reflectors and diffusers used in auditoria have been evaluated both objectively and subjectively. Two accurate systems have been developed to measure the scattering from surfaces via the cross correlation function. These have been used to measure the scattering from plane panels, curved panels and quadratic residue diffusers (QRDs). The scattering measurements have been used to test theoretical prediction methods based on the Helmholtz-Kirchhoff integral equation. Accurate prediction methods were found for all surfaces tested. The limitations of the more approximate methods have been defined. The assumptions behind Schroeder's design of the QRD have been tested and the local reacting admittance assumption found to be valid over a wide frequency range. It was found that the QRD only produces uniform scattering at low frequencies. For an on-axis source the scattering from a curved panel was as good as from a QRD. For an oblique source the QRD produced much more uniform scattering than the curved panel. The subjective measurements evaluated the smallest perceivable change in the early sound field, the part most influenced by reflectors and diffusers. A natural sounding simulation of a concert hall field within an anechoic chamber was used. Standard objective parameters were reasonable values when compared to values found in real halls and subjective preference measurements. A difference limen was measured for early lateral energy fraction (.048 ± .005); . inter aural cross correlation (.075 ± .008); clarity index (.67 ± .13 dB); and centre time (8.6 ± 1.6ms). It was found that; (i) when changes are made to diffusers and reflectors, changes in spatial impression will usually be larger than those in clarity; and (ii) acousticians can gain most by paying attention to lateral sound in auditoria. It was also found that: (i) diffuse reflections in the early sound field are not perceived differently from specular reflections; and (ii) the initial time delay gap is not significant to listener preference.