Pair-wise comparison experiment on subjective annoyance rating of noise samples with different frequency spectrums but same A-weighted level

Abstract A pair-wise comparison experiment on subjective annoyance rating of nine different noise samples is the focus of this paper. These noise samples collected from some associated facilities in urban residential areas are classified into three different types according to their frequency characteristics, and their A-weighted levels are equalized at 65 dBA. Based on the assessment of judging errors and statistical validity analysis of each testee’s evaluating matrix, invalid data are eliminated. The results show that the percentage of testees who chose noise samples containing more low frequency components had an obvious downside trend with the descending degree of annoyance from ‘the most annoyed’ to ‘the least annoyed’. The result indicates that noise which has low frequency characteristic is more annoying than noise of other frequencies when their A-weighted level are the same. Therefore, A-weighted level cannot scientifically assess noise annoyance which contains low frequency components mostly.