VR | 论文阅读
Three-dimensional virtual reality-based subjective evaluation of road traffic noise heard in urban high-rise residential buildings
基于三维虚拟现实的城市高层住宅道路交通噪声主观评价
Abstract
这篇文章提出了一种新的噪声评估方法，它应用以头部相关传递函数（HRTF）和头戴式显示器（HMD）为形式的虚拟现实（VR）技术来评估住宅内部的噪声。
HRTF是一种音效定位算法，提供方向信息，HMD提供视觉信息。
首先，在应用HRTF之前和之后，识别在起居室中记录的道路交通噪声的声压级和频率特性。其次，在四种不同测试环境中( without the HRTF or HMD, with the HRTF,with the HMD, and with both the HRTF and HMD)，对道路交通噪声为40-65 dB时的loudness, annoyance, disturbance, 和allowance进行主观反应评价。
HRTF提供的方向信息和HMD提供的视觉信息的影响分别为77％和23％。此外，基于对源和环境相关的空间属性的评估，HRTF对道路交通噪声的识别最高，并且HMD对空间的识别最高。当HRTF和HMD同时使用时，声音外化对VR环境的影响随着沉浸感和真实感的增强而增强。此外，噪声方向和噪声宽度的识别在心理声学影响中起重要作用，包括道路交通噪声的烦扰程度。
本研究的结果为外部噪声对室内环境的影响提供了新的见解。
1. Introduction
噪声会引起居民的反感，干扰他们的活动，甚至影响身体健康。
最近的研究表明，居民不仅受到噪声水平的影响，还会受环境因素的影响。先前的研究还表明，在评估环境噪声水平时观察到的视觉效果的变化会导致噪声感知的变化。例如，当显示植被覆盖的隔音墙的照片时，受试者对墙的偏好增加了，而当显示城市环境中的物体的照片并评估对噪声的响应时，主观反应取决于城市环境是否被辨认出。然而，这种评估存在一个问题，那就是它只是在屏幕上显示照片或二维图像，不能准确地反映对实际生活环境的感觉。尽管已经尝试通过家具布置和室内装饰来提供真实的测试环境，但是在重建具有不同条件的现实环境中仍然存在物理限制。
当引入增强现实（AR）来实现周围环境的视觉条件时，虚拟现实(VR)技术已经发展到非常接近现实的程度，该技术也正被用于对外部噪声的主观评估。这些测试能在根据噪声源的位置对其进行评估的同时，为处于虚拟模型空间中的参与者提供真实感。VR技术的优势在于能够在有限的实验室环境中模拟各种情况。
对HRTF的介绍：
对HMD的介绍：
本研究的目的是应用VR技术研究多层住宅建筑中的道路交通噪声，这是此类环境中最典型的外部噪声类型，以研究对噪声的主观响应。根据是否分别通过HRTF或HMD提供方向或视觉信息来评估各种噪声环境，以评估对进入建筑物的道路交通噪声，噪声源本身和空间环境的主观响应。
2. Methodology
2.1. Measurement
- measurements were taken in 5 min intervals on a weekday afternoon in
a standard-size living room in a high-rise residential building that faced
a busy eight-lane road.- a 1/2″ microphone (G.R.A.S) and an AS-70 portable sound level meter (RION) were placed at the center of the living room to obtain samples of the road traffic noise and record a single-channel sound source for use in the psycho-acoustic experiment.
- The road traffic noise was recorded a total of five times, during which the window was either opened or closed.
- The recorded sound sources showed a sound pressure level distribution in the range of Leq 38.8–55.5 dBA, and the sound pressure level was adjusted using the Amplitude function of the Audition software for auditory evaluation, as shown in Fig. 1.
- A total of six sound sources were constructed up to Leq 40.0–65.0 dBA.
2.2. Experimental design
2.2.1. Sound source and virtual room modeling
In this study, a CIPIC HRTF database containing the results of measurements of 43 people and two forms of ear on a KEMAR dummy head was employed.
Compared to the original sound, which is depicted in Fig. 1, a difference was detected between the sound pressure levels entering the two ears, while the frequency characteristics were similar, as shown in Fig. 2.
如图3所示，为了研究HMD对交通噪声的视觉影响，我们使用SketchUp 2018软件包[30]对实际记录声源的居住空间进行建模。利用Kubity软件对HMD模型进行可视化。在电视屏幕上合成一个视频，模拟测试对象坐在客厅里看电视，并对视频进行静音。
2.2.2. Experimental setup
To investigate the effects of the directional and visual information of the sound source on the road traffic noise entering the building, the noise was evaluated in four different test environments with different HRTF and HMD combinations, as summarized in Table 1.
During this test, the natural background noise of the experimental space was approximately 25 dBA.
In total, 12 sound sources were evaluated according to the sound pressure level of the road traffic noise before and after applying the HRTF.
2.3. Procedure
A total of 40 test subjects (28 males and 12 females) who had experienced road traffic noise while living in residential buildings participated in the evaluation.
实验对象在2秒间隔内两次暴露于12个声源中，其中包括6个不同噪音水平的单声道声源，分别代表进入大楼的道路交通噪音，以及6个应用了HRTF的声源。
此外，受试者还完成了一份问卷，以获取他们的震级估计值。每个测试对象评估所有四个不同的测试环境。
调查问题大致分为三类:1)对噪音的反应;2)对声源的感知;3)对空间的感知。
首先，采用7分制数值量表对道路交通噪声的响度、干扰度和活动干扰进行评价，以检验被试对交通噪声的主观反应。The allowance was defined on a three-point scale: “0: completely acceptable,”“1: somewhat acceptable,” and “2: definitely not acceptable.”声源的方向和宽度在7点数值尺度上进行评估，以确定对声源感知的差异。在相同的七分制量表上对沉浸感、真实感和外部化的质量进行了评估，以确定空间感知的差异
3. Results
3.1. Subjective response to road traffic noise
The subjective responses of the test subjects to the road traffic noise entering the building were investigated in terms of loudness, annoyance, and activity disturbance, and the results are presented in Fig. 4
The ratios of the annoyance level and activity disturbance versus LAeq changes in the four test environments are shown in Fig. 5 (a) and 5 (b).The percentage of annoyance means the ratio of subjects who evaluate the annoyance level > = ‘4: Moderate’.The percentages of participants who rated each noise level as “2: Definitely not acceptable” upon the occurrence of noise are shown in Fig. 5 ©.
采用Probit回归分析对评价环境的干扰程度和活动干扰进行校正。Probit regression
analysis involves a type of regression model that can analyze binomial response variables and transform the S-shaped dose response curve to a linear line that can be analyzed by regression through maximum likelihood.
3.2. Source-related spatial attributes
The direction and noise width associated with the recognition of the sound sources in the various environments were evaluated for each level of road traffic noise, and the results are presented in Fig. 6.
3.3. Environment-related spatial attributes
The difference in recognition in the spatial environments was investigated
through immersion, realism, and sound externalization, and the results are shown in Fig. 7.
4. Discussion
4.1. Impacts of audio-visual factors
Two-way ANOVA analysis was conducted to determine whether the
effects of the directional information (HRTF) and the visual information
(HMD) on the subjective evaluation were statistically significant. and
the results are summarized in Table 2.
4.2. Effects of spatial attributes on subjective responses
The correlation coefficients between the spatial attributes and
subjective evaluation results were calculated and are listed in Table 3,
all of which are statistically significant (p < 0.01).
The results of the regression analysis performed on the differences
between the spatial parameters and the subjective evaluation are presented
in Table 4.
Two-way ANOVA analysis was conducted to examine the effects of
the HRTF and HMD on the evaluation of the spatial attributes, and the
results are shown in Table 5.
However, because the HRTF and HMD interact with each other, as
indicated by the direction and realism evaluation results, the essential
primary effects of each variable were analyzed to characterize them,
and the results are shown in Table 6.