Abstract
Annoyance ratings obtained from listening experiments are widely used in studies on health effect of environmental noise. In listening experiments, participants usually give the annoyance rating of each noise sample according to its relative annoyance degree among all samples in the experimental sample set if there are no reference sound samples, which leads to poor comparability between experimental results obtained from different experimental sample sets. To solve this problem, this study proposed to add several pink noise samples with certain loudness levels into experimental sample sets as reference sound samples. On this basis, the standard curve between logarithmic mean annoyance and loudness level of pink noise was used to calibrate the experimental results and the calibration procedures were described in detail. Furthermore, as a case study, six different types of noise sample sets were selected to conduct listening experiments using this method to examine the applicability of it. Results showed that the differences in the annoyance ratings of each identical noise sample from different experimental sample sets were markedly decreased after calibration. The determination coefficient (R²) of linear fitting functions between psychoacoustic annoyance (PA) and mean annoyance (MA) of noise samples from different experimental sample sets increased obviously after calibration. The case study indicated that the method above is applicable to calibrating annoyance ratings obtained from different types of noise sample sets. After calibration, the comparability of annoyance ratings of noise samples from different experimental sample sets can be distinctly improved.