11.2 Noise Annoyance in Relation to Spatial Factors

This study examines effects of spatial factors extracted from the IACF on annoyance of noise stimuli in a laboratory. In the first experiment, using paired comparison tests the subjects judged their annoyance while changing fluctuations in the IACC and the SPL. In the second, they judged their annoyance while changing fluctuations in the τIACC and the SPL. Remarkable findings are that annoyance increased with increasing rates of the IACC and the τIACC fluctuations, in reference to increasing the SPL.

Environmental noises have been evaluated according to the SPL and their frequency characteristics. The noise criterion (NC) curve, the preferred noise criterion (PNC) curve, and the balanced noise criterion (NCB) curve have been developed for measurements of the SPL and its frequency characteristics (Beranek, 1957, 1971; Beranek et al., 1971; Beranek, 1988). Evaluations of temporal fluctuations involved in both traffic and industrial noise have used the equivalent sound level (Leq). The purpose of the current study, however, is to investigate effects of fluctuation of two spatial factors extracted from the IACF on annoyance of noise stimuli, which has not been known previously (Sato et al., 2004).

11.2.1 Experiment 1: Effects of SPL and IACC Fluctuations

Annoyance judgments were performed using paired comparisons while changing fluctuations in the interaural correlation magnitude IACC and sound pressure level SPL. The sound source was white noise. The rate of the IACC fluctuation was simulated by the frontal direct sound (L0) and two symmetric lateral reflections (L1 and L2) in a soundproof chamber. To produce an incoherent condition, the time delays between the direct sound and the two reflections were fixed at t1 = 20 ms andt2 = 40 ms. To change the IACC, a logarithmic envelope change was applied to the amplitude of the reflections. The period of the IACC change, PIACC, was set at 0.375, 0.75, 1.5 s, and ∞. The SPLs were set at 65, 70, and 75 dBA, and the resulting measured total SPLs were 65 ± 0.7, 70 ± 0.8, and 74.9 ± 0.8 dBA, respectively. Values of the running IACC measured with 2T = 0.1 s are shown in Fig. 11.8.

Fig. 11.8 Measured IACC for experiment 1. (a) PIACC = ∞. (b) PIACC = 1.50. (c) PIACC = 0.75. (d) PIACC = 0.375

When the IACC fluctuates, annoyance is assumed to be affected by the maximum value of its time gradient, which is defined by Flu_IACC, such that,

where is a small time interval. We have selected = 0.1 s obtaining its convergence in this study. Values of Flu_IACC that resulted were 2.34, 1.21, 0.72, and 0.17 for values of PIACC = 0.375, 0.75, 1.5 s, and ∞, respectively. Fluctuations in other spatial factors, τIACC and WIACC, could be fixed within –0.01 ± 0.01 ms and 0.02 ± 0.01 ms, respectively, because listeners always localized at the front in the median plane and the identical noise signal.

Paired comparisons tests were conducted for the 12 stimuli (four levels of Flu_IACC × three levels of SPL). Five subjects 22 to 24 years old with normal hearing ability participated in the experiment. A single test session consisted of 66 pairs (N(N – 1)/2, N = 12) of stimuli. Ten sessions were performed for each subject. A single test session was divided into two parts, each of which lasted 5.5 min, to prevent subject fatigue. Fifty responses (5 subjects × 10 sessions) to each stimulus were obtained. Consistency tests indicated that all subjects had a significant (p < 0.05) ability to distinguish various degrees of annoyance.

As shown in Fig. 11.9, the scale value of annoyance increased as the Flu_IACC and the SPL increased. A Flu_IACC > 0 was consistently judged to be more annoying than the condition of the Flu_IACC = 0 when the SPL was constant. Results of the analysis of variance for scale values of annoyance indicate that the factors Flu_IACC and the SPL are significant (p < 0.01). Effects of the interaction between Flu_IACC and the SPL were not significant, so that the Flu_IACC and the SPL contributed to the scale value of annoyance independently, thus,

Fig. 11.9 Average scale value of annoyance as a function of Flu_IACC and as a parameter of SPL. •, SPL = 65 dBA; , SPL = 70 dBA; and , SPL = 75 dBA

where coefficients a ≈ 0.46 and b ≈ 0.20 were obtained by multiple regressions. It is remarkable that the calculated scale values agree well with the measured ones with a correlation coefficient of 0.99 (p < 0.01).

For each individual, weighting coefficients a and b in Equation (11.3) were also obtained (Table 11.3). Two of five subjects indicated that the scale value did not satisfy the model’s goodness of fit, because the probability of their judgments in the PCT was beyond the linear range of the scale value (0.05 < p < 0.95). Therefore, coefficients in Equation (11.3) for subjects A and D could not be obtained.

Table 11.3 Contributions and coefficients in Equation (11.3) for each individual in experiment 1

11.2.2 Experiment 2: Effects of Sound Movement

Annoyance judgments were performed using paired comparison tests while changing fluctuations in the τIACC and the SPL. The source signal was the band-pass filtered noise with a center frequency of 500 Hz (bandwidth = 160 Hz). The horizontally swaying sound images were simulated by the two lateral, symmetric loudspeakers (L1 and L2) in the soundproof chamber. In these experiments, the horizontal angles of lateral sounds were ±54◦.

To produce the sound field with the fluctuation of τIACC, the logarithmic envelope was applied to the amplitude of noise. The period PτIACC was set at 0.375, 0.75, 1.5 s, and ∞. The SPL values were set at 65, 70, and 75 dBA and the ranges of the measured total SPL were 64.8 ± 1.5, 69.6 ± 1.6, and 74.7 ± 1.5 dBA, respectively. The measured values of τIACC are shown in Fig. 11.10. To follow the fluctuations of τIACC, the IACF was calculated for the integration intervals 2T = 0.3 s.

When the value of τIACC fluctuates, then annoyance can be affected by the maximum value of its time gradient, which is defined by Flu_τ IACC

Here, is selected 0.3 s small enough for obtaining the convergence. Values of Flu_τ IACC obtained were 3.73, 2.07, 1.50, and 0.07 for PIACC = 0.375, 0.75, 1.5 s, and ∞, respectively. Fluctuations of the measured IACC and WIACC were almost constant within 0.91 ± 0.08 ms and 0.27 ± 0.01 ms, respectively.

Fig. 11.10 Measured τIACC for experiment 2. (a) PτIACC = ∞. (b) PτIACC = 1.50. (c) PτIACC = 0.75. (d) PτIACC = 0.375

Paired comparisons tests were conducted for 12 stimuli (four levels of Flu_τ IACC × three levels of SPL). The same five subjects from experiment 1 participated. They were asked to judge which of two stimuli they perceived to be more annoying. The duration of the stimuli was 3 s, the rise and fall times were 50 ms, and the silent interval between the stimuli was 1 s. Each pair of stimuli was separated by an interval of 3 s and presented randomly. A single test session consisted of 66 pairs (N(N – 1)/2, N = 12) of stimuli. Ten sessions were performed for each subject. A single test session was divided into two parts, each of which lasted 5.5 min. Fifty responses (5 subjects × 10 sessions) to each stimulus were obtained.

Figure 11.11 shows that the scale values of annoyance increased as Flu_τ IACC and the SPL increased. When the SPL was constant, the moving sound images (Flu_τ IACC > 0) were always more annoying than the fixed sound image (Flu_τ IACC = 0). Results of an analysis of variance for scale value of annoyance indicate that the factors Flu_τ IACC and SPL are significant (p < 0.01). Effects of the interaction between Flu_τ IACC and SPL are not significant. Thus, Flu_τ IACC and the SPL contribute to the scale value of annoyance independently, so that:

Fig. 11.11 Average scale value of annoyance as a function of Flu_τ IACC and as a parameter of SPL. •, SPL = 65 dBA; , SPL = 70 dBA; and , SPL = 75 dBA

where coefficients a ≈ 0.42 and b ≈ 0.16 were obtained by multiple regression. The calculated scale values agree well with the measured ones with a correlation coefficient of 0.98 (p < 0.01).

As the contribution of the factors Flu_τ IACC and the SPL to the scale value of annoyance for each individual, coefficients a and b in Equation (11.5) were also obtained (Table 11.4). One of five subjects indicated that the scale values did not satisfy the model of goodness of fit. Therefore, the coefficients given in Equation (11.5) for subject E were not obtained. A comparison of the results from experiment 1, for all subjects, shows the effects of the SPL are relatively small, and Flu_τ IACC cannot be ignored for the evaluation of annoyance. As listed in Table 11.4, the contribution of the Flu_τ IACC to the scale value of annoyance apparently was greater than that of Flu_IACC. It is noteworthy that the variations in the ranges of SPL were the same from 65 to 75 dBA in both experiments. Because τIACC = 0 is one of the preferred conditions for listening in the sound field (see Section 3.2.4), annoyance increased with the stimuli of Flu_τ IACC > 0.

Table 11.4 Contributions and coefficients in Equation (11.5) for each individual in experiment 2

The value of the IACC was between 0.2 and 0.8 in experiment 1, however, subjects did not perceive a fluctuation in the IACC when the IACC was such a low value (see, e.g., Section 7.3). Or, the just noticeable difference (JND) of the IACC in the sound field with a lower IACC is larger than that with a higher IACC (Gabriel and Colburn, 1981). In addition, the sound image is blurred when the IACC is low.

On the other hand, in experiment 2, the value of τIACC was changed from –0.4 to 0.4 ms, and PτIACC was 0.38 s at the minimum. The threshold of the interaural time delay (for the 1000-Hz tone) is short enough, 10 μs (Klumpp and Eady, 1956). The duration of 0.38 s is long enough to perceive the movement of the sound source (Grantham, 1986, Chandler and Grantham, 1992). In addition, all stimuli had clear sound image, because the values of the IACC for all stimuli in this experiment were greater than 0.82. Therefore, subjects perceived the fluctuation of localization according to τIACC during the whole period of the stimuli much more vividly than the condition of the IACC fluctuation. Under two experimental conditions, 3.73 of the Flu_IACC is equivalent to an increase of 4.9 dB in the SPL, and 2.96 of the Flu_τ IACC is equivalent to an increase of 9.7 dB in the SPL.