close

Вход

Log in using OpenID

Effects of active noise control on subjective annoyance and cortical

embedDownload
Effects of active noise control on subjective annoyance and cortical
neural activities for car engine noise
Tomoki Ito1,2, Shunsuke Ishimitsu1, and Seiji Nakagawa2
1
Hiroshima City University
3-4-1 Ozuka-Higashi, Asa-Minami, Hiroshima 731-3194 JAPAN
2
National Institute of Advanced Industrial Science and Technology (AIST)
1-8-31 Midorigaoka, Ikeda, Osaka 563-8577, Japan
ABSTRACT
It is thought that active noise control (ANC) is able to control not only loudness but also auditory
impression; however, few studies have focused on changes in the auditory impressions of noises
through ANC. Some previous studies have reported significant correlations between subjective
preference/annoyance of sounds and synchrony of the alpha activity, e. In this study, auditory
impressions of car engine noises varied by ANC were investigated using psychoacoustic and
neurophysiological measurements. Car engine noises were modeled as harmonic complex tones. 20
stimuli were obtained by applying ANCs with several reduction-levels. First, Scheffe’s paired
comparison tests were performed to evaluate subjective annoyance for each stimulus. Next,
magnetoencephalographic responses for the reference noise, the intermediate -, and the least-annoyed
stimuli for each participant were recorded, and then e was estimated accordingly. Annoyance
decreased as the cut-off frequency and the reduction-level increased. The results indicated that
decrease of loudness is important for improvements in auditory impressions. However, no significant
effects of ANC were observed in e. It is difficult to identify the cause, the annoyance levels for each
stimulus were not markedly different.
Keywords: ANC, paired comparison test, magnetoencephalography, alpha wave
1. INTRODUCTION
Active noise control (ANC) is a noise-reduction technique that works by the addition of one or
several secondary sources to the original sound source. ANC is especially suitable for reducing
low-frequency noise where traditional passive noise control techniques based on sound -absorbent
materials and acoustic screens are ineffective. Therefore, ANC techniques are widely used for cars,
headphones, and so on. It is thought that ANC reduces noise loudness, but also changes auditory
impressions. However, few studies have focused on the quantitative effects of ANC on auditory
impressions. Gonzalez et al found that auditory impressions were improved by applying ANC to
engine noises (1). However, in the case of smoother spectrums, auditory impressions were
deteriorated. They suggested the possibility that auditory impressions are affected when the
psychoacoustic parameters of expecting loudness are varied. Moreover, ANC is more likely to
deteriorate auditory impressions if the original source sound has positive impressions. In accelerating
cars, for example, the time-varying characteristics of the engine sound can induce auditory
impressions such as “sporty” and “accelerating” in drivers. Conversely, assessment of sound quality is
carried out using subjective psychological tests such as the paired comparison me thod and semantic
differential method. However, such psychoacoustic tests usually compel participants to repeat
judgments over and over, and often take a long time. Furthermore, it is sometimes hard to obtain an
entirely reliable evaluation. It is comparatively easy to improve the sound quality if we can also use an
objective measure corresponding to the auditory impression. It is thought that usage of
1
t-ito@hfce.info.hiroshima-cu.ac.jp
Inter-noise 2014
Page 1 of 7
Page 2 of 7
Inter-noise 2014
neurophysiological data is one of the most reliable methods. We previously examined to reveal the
relationships between the brain cortical neural activities and subjective preference or annoyance of
sounds (2-4). In these studies, magnetoencephalographic (MEG) components in the 8–13 Hz range, i.e.,
alpha activities, were recorded and analyzed using an auto-correlation function (ACF). It was found
that the effective duration of using the ACF, e, which is related to repetitive features within the signal
itself, was lengthened for the preferred stimuli (2, 3) and shortened for the annoyed stimuli (4). In this
study, the effects of ANC on the auditory impression of car-engine noise were investigated using both
subjective psychological and neurophysiological measurements. First, the scale values of annoyance
for each participant were obtained by paired-comparison tests. Next, MEG measurements and the ACF
analyses were conducted. The relationship between the scale value for annoyance of the car engine
noise and the factors extracted from the ACF in the brain’s magnetic responses were investigated.
2. STIMULI
In this study, we used a 2.3L inline-four engine. The engine condition was set to full constant speed
in third gear. Figure 1 shows a spectrogram of an inline-four engine sound, recorded at a sampling
frequency of 6000 Hz with a dummy head located on the passenger sea t. An inline-four engine
explodes twice with each rotation and generates a secondary vibration force. These higher harmonics
constitute the main component of the noise. The modeled stimulus was generated based on a recorded
sound. We used a sinusoidal model to create the stimulus (5). The model implements analysis and
synthesis for sound waves, as shown in Fig. 2. The peaks of the spectrogram in each frame were
extracted using a short-term Fourier transform (STFT), and partial sounds were determined by the
instantaneous values included in their peaks.
Figure 1 - Spectrogram of engine noise recorded by a dummy head located on the passenger seat.
Figure 2 - Analysis-by-synthesis with Sinusoidal Model
Page 2 of 7
Inter-noise 2014
Inter-noise 2014
Page 3 of 7
Reference noises were created from the real car engine noise using the sinusoidal model. Modeled
car engine noise means the harmonics up to the 42nd based on a real inline-four engine noise.
Random phase noise means the collapsed phase of the modeled car engine noise. To make the
A-weighted sound pressure level (LA) equal to the level of the recorded real car sound, the
amplitude of the reference noises were adjusted. Moreover, one of the following actions was
performed on these stimuli. For ideal-reduction noise, four stimuli had their amplitudes reduced by 5,
10, 15 or 20 dB. For modeled ANC noise, low frequency parts of the reference noises were reduced.
The cut-off frequencies and levels of reduction were set at 200, 400 or 800 Hz, and 5, 10, 15 or 20
dB, respectively. To evaluate equal LA noise, i.e., the equal L A of the ideal-reduction noise, the levels
of reduction in each frequency part were adjusted between 0 and 20 dB.
3. SUBJECTIVE ANNOYANCE MEASUREMENT
3.1 Participants
Ten volunteers (18–25 years old; seven male, three female) with normal hearing participated in this
experiment. They were seated in a soundproof room with a comfortable thermal environment.
3.2 Apparatus and tasks
Scheffe’s paired-comparison tests were performed for all combinations of the pairs of stimulus for
each reference noise, i.e., 190 pairs (N(N−1)/2, N=20) of stimuli with a different order of each pair per
session, and random presentation of the pairs. Each pair of stimuli was presented four times to
determine the annoyance scale value. Pairs of sounds were presented, and participants were asked to
judge their annoyance at the constant speed car engine noise to estimate the annoyance scale value on
a scale of 1 to 7. These auditory stimuli were presented through plastic tubes and earpieces (ER -2,
Etymotic Research) inserted into the ear canals. The scale values of the annoyance were calculated
according to Scheffe’s theory.
3.3 Results
Figure 3(a) and (b) shows the relationship between the frequency width and intensity of active
noise control and the scale values of the annoyance for the modeled ANC n oise and the equal L A
noise in the modeled car engine noise, respectively. When the level of the ideal-reduction noise
increased, the annoyance was decreased from the reference noise, as shown in Figure 3(a).
Furthermore, the width and level of the modeled ANC noise increased, while the scale values of the
annoyance decreased. In comparing the equal L A noise, the annoyance of the modeled ANC noise
was smaller than the reduction level of the ideal-reduction noise, as shown in Figure 3(b). However,
the frequency width of the modeled ANC noise increased, and the annoyance was closer in the case
of the ideal-reduction noise. Figure 4(a) and (b) shows the relationship between the frequency and
the intensity of active noise control and the scale values of the anno yance for the modeled ANC
noise, the equal L A noise and the random phase noise. In the random phase noise, the result showed
the same tendency as the modeled ANC noise.
(a) Reference noise
Inter-noise 2014
Page 3 of 7
Page 4 of 7
Inter-noise 2014
(b) Equal LA noise
FIGURE 3. Scale values of annoyance and frequency width and intensity of active noise control in
the modeled car engine noise.
(a) Reference noise
(b) Equal LA noise
FIGURE 4. Scale values of annoyance and frequency width and intensity of active noise control in
the random phase noise.
Page 4 of 7
Inter-noise 2014
Inter-noise 2014
Page 5 of 7
4. MEG MEASUREMENT
4.1 Participants
The same volunteers used in the Scheffe’s paired comparison tests participated in the recording of
MEG responses. They were seated in a magnetically-shielded room with a comfortable thermal
environment. The auditory stimuli were presented using the same earpieces as in the paired
comparison test (ER-2, Etymotic Research).
4.2 Tasks
According to the annoyance levels of Scheffe’s paired comparison test, the reference noise, the
intermediate- and the least-annoyed stimuli in the modeled ANC noise were decided for each
participant and each reference noise. Participants were requested to carry out Thurston’s
paired-comparison tests for all combinations of stimuli for each reference noise, i.e., three pairs
(N(N−1)/2, N=3) of stimuli with a different order of each pair per session, and random presentation
of the pairs. They were asked to judge the annoyance of the constant speed car engine noise, and to
close their eyes.
4.3 MEG recording and analysis
MEGs were recorded with a 122-channel whole-head MEG system (Neuromag-122 TM , Neuromag
Ltd.) in a magnetically shielded room in AIST Kansai Center. MEG recordings were conducted at the
same time as the Thurston's paired-comparison tests. The effective durations of autocorrelation
functions (ACF) of the alpha wave, e, previously reported as an objective index of
preference/annoyance, were obtained (2-4). In this study, the alpha wave was extracted by bandpass
filtering between 8 and 13 Hz. Autocorrelation functions (ACF) of the alpha wave were calculated
for each stimulus. Figure 5 shows an example of a recorded MEG between 8 and 13 Hz. We defined
three channel sets, the left and right temporal and occipital areas, as shown in Fig. 5. ACF of the
alpha wave and its effective duration, e, were calculated using the same method as previous study
[5].
FIGURE 5. An example of recorded MEG alpha activities between 8 and 13 Hz.
Three channel sets, the left and right temporal, and the occipital areas, were analyzed.
4.4 Results
Figure 6(a) and (b) show the relationships between the scale values of subjective annoyance and the
e, obtained from the three channel sets in the modeled car engine noise and the random phase noise,
respectively. The subjective annoyance of stimuli had no significant effect on e for these noises.
Inter-noise 2014
Page 5 of 7
Page 6 of 7
Inter-noise 2014
(a) Modeled car engine noise
(b) Random phase noise
FIGURE 6. Relationships between the scale values of subjective annoyance and values of e.
5. DISCUSSION
In the subjective annoyance measurement, the annoyance of reference noises was the highest in
most participants. When the cut-off frequency and reduction-level of the modeled ANC noise
increased, the annoyance decreased. Generally, the sharpness of the stimulus was increased by ANC.
The results indicated that the decrease in loudness was more effecti ve than for the equal LA noise;
when the cut-off frequency of ANC was increased, the annoyance was decreased. We came to the
conclusion that the effectiveness of reduction in the sharpness was perceived owing to the high
frequency parts. In the MEG measurement, unlike the previous study, the subjective annoyance of
stimuli had no significant effect on the duration of the alpha wave. As the car engine noise is
considered an active sound, it is possible that the impression was not reflected the relaxed state to
the alpha wave. The effectiveness of ANC was confirmed for constant speed engine noise, as in
random phase noise. Further investigations of other vehicle operations, such as acceleration, are topics
for future research.
6. CONCLUSION
Subjective annoyance decreased as cut-off frequency and reduction-levels of ANC increased.
Moreover, in the equal L A noise, when the cut-off frequency of the modeled ANC noise increased, the
annoyance was closer to the ideal-reduction noise. Conversely, no significant effects of ANC were
observed in e of the alpha activity in the MEG measurement. In the random phase noise, the result
Page 6 of 7
Inter-noise 2014
Inter-noise 2014
Page 7 of 7
showed the same tendency as the modeled ANC noise in both the results of subjective annoyance
measurement and MEG measurement. These results indicated that a decrease in loudness is important
for improvements in the auditory impression of constant speed car engine noises.
7. REFERENCES
1. A. Gonzalez, M. Ferrer, M de Diego, G. Pinero, J.J. Garcia-Bonito, ” Sound quality of low-frequency
and car engine noises after active noise control”, Journal of Sound and Vibration , 265, 663–679, 2003.
2. Y. Soeta, S. Nakagawa, M. Tonoike and Y. Ando, Magnetoencephalographic responses corresponding to
individual subjective preference of sound fields, Journal of Sound and Vibration, 258, 419-428, 2002.
3. Y. Soeta, S. Nakagawa and M. Tonoike, Magnetoencephalographic activities related to the magnitude
of the interaural cross-correlation function (IACC) of sound fields, Journal of Temporal Design in
Architecture and Environment, 5, 5-11, 2005.
4. Y. Soeta, S. Nakagawa, M. Tonoike and Y. Ando, “Magnetoencephalographic responses correspond to
individual annoyance of bandpass noise”, Journal of Sound and Vibration, 277, 479-489, 2004.
5. S. Ishimitsu, H. Nishikawa, T. Hotehama, Y. Soeta, S. Nakagawa, “Measurement of brain magnetic
fields corresponding to subjective preference of car additional sound”, Proceeding of the 2010
symposium of Environmental Engineering, 20, 73-75, 2010.[In Japanese]
Inter-noise 2014
Page 7 of 7
1/--pages
Пожаловаться на содержимое документа