Abstract:
In preliminary studies [May et al., Proc. INTER-NOISE 96 (1996)] it was found that subjective human response (annoyance) to refrigerator noise could be predicted well using a model based on three objective signal metrics. The first study involved 24 subjects, half with a Korean background and half with an American background, and 17 refrigerator signals. The subjects of Korean background exhibited a heightened sensitivity to the noises. A second study used 24 subjects, all with an American background, and 12 signals. Paired comparison and parametric tests produced similar results. A third study investigated the feasibility of deriving a general purpose objective model to predict annoyance due to steady-state refrigerator noise. A larger range of signals was used to better represent refrigerator noises likely to be encountered in practice. The increased number of signals made paired comparison tests impractical and thus only the similarly effective parametric tests were performed. Linear combinations of signal metrics yielded models with R[sup 2] values higher than 0.8, indicating a high degree of predictability. [Work supported by LG Electronics and GE Appliances.]