The purpose of the present paper is to clarify the effect of sound duration on annoyance, i. e. how the duration effect (annoyance increase per doubling of duration) can be quantified, and whether or not the duration effect depends on the sound level (level-dependent) and/or the length of duration (duration-dependent). Stimulus sound is white noise whose duration ranged from 30 ms to 90 s for each of four levels. Idealized time patterns of signal employed in the judgment test are shown in Fig. 3, and some physical characteristics of the stimulus sounds are presented in Table 1. In order to avoid too long time for one test, all stimulus sounds were divided into three groups, I, II and III, as shown in the table. The data obtained in these experiments can be compared with each other through common stimuli (marked with asterisk). Ten male and ten female students with normal hearing acuity served as subjects. They were given abundant practice before the actual experiment began. The subjects were instructed to judge the whole perceived magnitude of the sound, for example, annoyance, unpleasantness, etc. The sounds were arranged at random with respect to test variables and presented to each subject through a headphone (diotic) in the sound proof room. The subjects assigned numbers proportional to the magnitudes of his perception for the sounds presented one after another. The number denoting annoyance of a sound, the level and the duration of which were 80 dB and 10 s respectively, was arbitrarily equated to one and the other annoyance estimates for each subject were transformed with reference to this value. According to the results shown in Fig. 5, the annoyance increases linearly with the logarithm of the length of duration. The following equation was obtained by multiple regression analysis with two independent variables: logψ=0. 229 logt +0. 0203L-1. 842, where ψ is the ratio of annoyance, t is the length of duration (s), and L is the peak level (dB). From the equation, it can be seen that annoyance becomes twice when the level of sound increases by 14. 8 dB. Using this relation, ψ is expressed as the relative sound pressure level (RSPL), which is shown in the left ordinate of Fig. 5. Taking the interaction between L and t into account, we obtain the following equation: logψ=0. 0018L logt +0. 091 logt +0. 020L-1. 817. Since all regression coefficients in this equation are significant (significant level; 0. 05), the duration effect is considered to be level-dependent. By expressing L as a function of t and ψ in the equation, we obtain equal annoyance contours for some values of ψ (Fig. 10), which show hyperbolic curves because of the interaction term. Therefore, the apparent duration-dependency of duration effect observed in Fig. 10 should be attributed to its level-dependency. Experimental results are also plotted (Fig. 11) against acoustic energy level of the stimulus sound (L_<en> in Eq. (16)). It can be seen that the annoyance increases approximately linearly with L_<en>.
抄録全体を表示