Synergetic effect of vehicle interior sound and design on comfortability in cabin Junji Yoshida 1 Kanta Imamori 1 Osaka Institute of Technology 5-16-1 Omiya, Asahi-Ku, Osaka, 535-8585, Japan

ABSTRACT Vehicle interior noise have been quiet by the development of noise and vibration reduction technique. This noise will become quieter by the engine rest in electric vehicle. Hence, vehicle occupants may expect much more comfort in cabin as if they are in living room. In addition, not only the auditory stimulus but also visual stimulus is considered to have important role for the comfortability in cabin. In this study, we then focused on the comfortability in cabin and investigated influence of vehicle interior sound and design on comfortability using VR system. In the test, various virtual interior design and interior sound recorded binaurally at cruising condition were presented in the VR system and the sound, design and total comfortability was assessed through subjective evaluation test. As the result, both sound and design comfort were observed to be essential to increase the total comfort- ability and synergetic effect between them was found. This indicates that the influence of sound com- fort to total comfortability changes according to the design comfort. Furthermore, soft-smooth inte- rior sound and dark-solid interior design were clarified to improve the total comfort.

1. INTRODUCTION

In near future, vehicle interior expects to be more comfortable due to the progress of electrical and autonomous vehicle technology. Interior noise is very important factor of the comfort in cabin. Hence, a lot of activities for evaluating or improving interior noise have been carried out [1],[2]. On the other hand, when we drive a vehicle and hear interior noise, not only the auditory information but also visual and vibration information are given. Accordingly, some researches focused on the interaction of the multi-sensors using driving simulator [3]-[6]. In this study, we then focused on the influence of vehicle interior sound and design on the total comfortability in cabin using virtual reality (VR) system. Various interior sound and design were prepared for the VR system in the subjective evalu- ation test, and the influence of them was analyzed from the evaluation results. In addition, we at- tempted to propose interior sound and design characteristics to realize the comfort cabin by making total interior comfort evaluation model using the interior sound and design characteristics quantita- tively. 2. SUBJECTIVE EVALUATION TEST

We carried out three subjective evaluation tests about the comfortability in cabin using VR system. In the first test, we asked the sound quality to the presented vehicle cruising interior sound. The comfortability of interior design was asked in the second test. And in the third test, the total comfort- ability of the interior sound and design were assessed.

1 junji.yoshida@oit.ac.jp

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

2.1. Experimental System and Evaluation Terms In this experiment, the subjective evaluation system employing VR system was prepared as shown in Figure 1. Participants sat on the vehicle seat and evaluated interior sound and design. The interior sound was reproduced binaurally via headphones (SENNHEISER HD600) from play-back system (HEAD acoustics SQobold). The interior image was projected via head-set (HTC VIVE Pro Eye) using VR system (Valve Corporation Steam VR). The participants sitting on the passenger seat looked at the interior design as they like by moving their head during the test.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Figure 1: Experimental apparatus using VR and playback systems.

Automobile seat Playback system (SQobold)

In the first evaluation test (TEST 1), only interior sound was presented and the participants evalu- ated the comfort of the presented vehicle cruising interior sound by selecting one of the following five choices (5: Very comfort, 4: Comfort, 3: Neither, 2: Discomfort, 1: Very discomfort). In addition, “Loud”, “Pitch”, and “Rough” of the presented sound were also asked to investigate factors affecting the sound comfort. In the second evaluation test (TEST 2), only interior design was presented and they evaluated the comfort of the design in addition to the evaluation of “Brightness”, “Open feeling”, “Soft” as the same manner of TEST 1. In the third test (TEST 3), both interior sound and design were presented and they evaluated only comfort of the total interior condition.

2.2 Presented Stimuli and Evaluation Procedure Ten vehicle cruising interior sounds (CS1 – CS10) and four interior design (ID1 – ID4) were prepared in the subjective evaluation test to investigate the comfort in cabin. The interior sounds were recorded binaurally and the A-weighted SPL were from 59 to 67 dB. Figure 2 shows a part of employed four vehicle interior images in the VR system. As shown in this figure, instrumental panel and center console was separated in ID1 and 2 design images. And they were connected in ID3 and 4 images. About the color, ID1 and 3 were dark colored and ID2 was pastel and ID4 was beige colored design. About the presented visual information unrelated to the vehicle, white lines on the road and simple buildings were prepared in TEST 2 and 3. In TEST 2, the vehicle stopped, hence the lines and build- ings did not move but this information moved according to the vehicle speed in TEST 3. And partic- ipants could move their heads to look around in cabin during the test.

(a) ID1. (b) ID2. (c) ID3. (d) ID4. Figure 2: Presented interior design images in VR system.

After hearing or looking at the vehicle interior sound and design, the participants chose appropriate category in the five categories in each test. 2.3 Participants Thirteen male and female in their 20’s and 30’s participated in this test. We obtained the informed consent accepted by the ethic committee in our university from all participants. Three of them drove vehicles one in a week, and three drove ones in a month, and four of them drove rarely (once in a half year) and the rest (three) did not have driving experience. About the number of evaluation, there were 10 sounds and 4 interior design images in TEST 1 and TEST 2, respectively. Accordingly, there were 40 patterns (10 sounds x 4 images) in TEST 3 in which both sound and image were presented.

The test was carried out in session. In each session, there were TEST 1 (10 sounds evaluation), TEST 2 (4 images evaluation), and TEST 3 (40 patterns evaluation). And each participant performed total three sessions. The presentation order was randomized to prevent from the order influence. Therefore, total sound evaluation number became 390 (13 participants x 10 sounds x 3 sessions). And total interior design evaluation number became 156 (13 participants x 4 images x 3 sessions). About the number of combination evaluation in TEST 3 was 1560 (13 participants x 40 combinations x 3 sessions). Accordingly, each sound or visual or combination stimulus was evaluated 39 times (13 participants x 3 sessions) in total in TEST 1, 2 and 3, respectively. 2.4 Subjective Evaluation Result Figure 3 (a) shows the averaged comfort score of each presented sound obtained from TEST 1. Error bar shows the standard deviation.

5

5

Good

Good

Interior design comfort

4

4

Sound comfort

3

3

2

2

1

CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 Bad

1

Bad

ID1 ID2 ID3 ID4 Image number

Sound number

(a) Comfort score of interior sound (b) Comfort score of interior design

Figure 3: Comfort score of interior sound and design. As shown in the upper figure, CS6 and CS8 are observed to be evaluated as comfort among 10 sounds. Figure 3 (b) shows the averaged comfort score of each presented interior design image obtained from TEST 2. Black, light blue, dark gray and orange bars show the comfort score of ID1, 2, 3 and 4, respectively. As shown in this figure, the dark colored interior designs (ID1 and 3) were evaluated to be relatively higher comfort. Figure 4 shows the total comfort score to the sound and design combination. In the figure, black, light blue, dark gray, and orange bars show the comfort score of ID1, 2, 3 and 4, respectively. As shown in the figure, comfort score of CS6 and CS8 were evaluated to be comfort as same as TEST 1 (sound evaluation). This illustrates that the influence of sound evaluation was high on the total comfort evaluation. On the other hand, the total comfort of ID1 and ID3 were observed to be relatively higher than that of ID2 and 4 in each presented sound group. This also shows that the interior design affects the total comfort. However, the design influence differs in each sound group,

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

i.e., the influence in CS6 and CS8 (comfortable sound group) was large but the influence in low comfort sound group (CS1 and CS5) was small. From these results, the synergy effect between sound and interior design image may exist in the evaluation of the total comfort in cabin.

Good

5

ID1 ID2 ID3 ID4

4

Total comfort

3

2

1

Bad

CS1 CS2 CS3 CS4 CS5 CS6 CS7 CS8 CS9 CS10 Sound number

Figure 4: Total comfort score to each presented interior sound and design image combination. 2.5 Influence of Vehicle Interior Sound and Design on Comfort in Cabin In this section, we analyzed the influence of vehicle interior sound and design on total comfort in cabin by using the three subjective evaluation results. For obtaining the influence, multiple regression analysis was applied. The evaluation result of TEST 1 (sound comfort score) and the result of TEST 2 (design comfort score) were used as the explanatory variable and the result of TEST 3 (total comfort score) was used as the response variable. Here, the synergy effect was not included in this regression analysis. Equation 1 shows the obtained regression equation. The regression coefficient of each ex- planatory variable is standardized coefficient.

Total comfort = 0.85 × Sound comfort + 0.45 × Design comfort (1)

As shown in this equation, the coefficient of sound comfort is almost twice of the design comfort coefficient. This shows that the influence of sound was much larger than that of design comfort. Figure 5 (a) shows the relationship between subjective total comfort score obtained from TEST 3 and calculated total subjective score by using Equation 1.

5

Good

5

Good

4

4

Total comfort

Total comfort

Subjective

Subjective

3

3

2

2

R = 0.959

R = 0.967 Bad

1

1

1 2 3 4 5 Bad

1 2 3 4 5

Bad Estimated Total comfort Good

Bad Estimated Total comfort Good

(a) Without synergy effect (b) With synergy effect Figure 5: Relationship between subjective and calculated total comfort score.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

As shown in the figure, the correlation coefficient was very high (0.94) and the equation was found to express the total comfort very well. Subsequently, multiple regression analysis was again carried out to consider the synergy effect between sound and design comfort. For considering the synergy effect, we prepared three explanatory variables (sound comfort score, design comfort score, and sound comfort score x design comfort score) and stepwise method was applied to find the significant explanatory variable. As the result, only synergy effect item (sound comfort score x design comfort score) was selected in the regression equation as shown in Equation 2.

Total comfort = 0.97 × Sound comfort score × Design comfort score (2)

Figure 5 (b) shows the relationship between subjective total comfort and calculated total comfort considering the synergy effect by Equation 2. As shown in the figure, the correlation coefficient was very high and the equation could express the subjective total comfort very well. In addition, the cor- relation coefficient was increased by considering the synergy effect as shown in Figure 5 (a) and (b). From these results, it was found that the influence of sound and design comfort is considered to affect each other and in case the design (/sound) comfort is large, the influence of the sound (/design) com- fort increases. 3. QUANTIFICATION OF TOTAL COMFORT IN CABIN

3.1 Factor Analysis of Sound Comfort In this section, for understanding what makes comfort sound in cabin, we carried out regression anal- ysis again. In this part, only sound evaluation results in TEST 1 was used. As the explanatory variable, loud score, pitch score and rough score to the presented cruising sound were used. And sound comfort was employed as the response variable. Equation 3 shows the regression equation.

Sound comfort = - 0.52 × Loud - 0.53 × Rough (3)

As shown in this equation, the sound comfort could be express using only loud and rough of the sound. And the correlation coefficient between subjective sound comfort score and calculated sound comfort score using Equation 3 was very high at 0.98. Hence, soft and smooth sound was found to be important to make comfort interior sound.

In the next step, to estimate the sound comfort from the recorded sound pressure signal, psychoa- coustic metrics were introduced. Loudness calculated from sound pressure signal indicated high cor- relation with subjective loud score, and partial roughness under 10 Bark was high correlation with subjective rough score [8]. Then, the sound comfort equation was re-built by using the psychoacoustic metrics as shown in Equation 4. Figure 6 shows the relationship between subjective sound comfort score and calculated sound comfort score by using Equation 4.

Sound comfort = 10.87 - 0.19 × Loudness - 25.85 × Partial Roughness (under 10 Bark) (4) From these results, the sound comfort could be expressed very well from the recorded sound pres- sure signal by using psychoacoustic metrics quantitatively.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

5

Good

4

Total comfort

Subjective

3

2

R = 0.933 Bad

1

1 2 3 4 5

Bad Estimated Total comfort Good

Figure 6: Relationship between subjective and calculated sound comfort score

using psychoacoustic metrics. 3.2 Factor Analysis of Interior Design Comfort and Quantification of Total Comfort In this part, we attempted to find factors affecting the interior design comfort (TEST 2). Here, it was hard to make quantitative evaluation model as sound comfort evaluation model because the sample number of interior design was four in spite of a lot of information in the interior design. Hence, we evaluated the interior design characteristics correlating the design comfort. Figure 7 shows the aver- age score of each evaluation term (comfort, brightness, open feeling, and soft).

Good

5

Bright

5

4

4

Brightness

Comfort

3

3

2

2

Bad

1

1

Dark

ID1 ID2 ID3 ID4 Image number

ID1 ID2 ID3 ID4 Image number

Open

5

5

Soft

4

4

Open feeling

Soft

3

3

2

2

1

1

Solid

Close

ID1 ID2 ID3 ID4 Image number

ID1 ID2 ID3 ID4 Image number

Figure 7: Evaluation score to interior design images in each term.

As shown in these figures, the design comfort of ID1 and ID3 were high but the comfort of ID2 was low. This tendency was opposite to the other factors (brightness, open feeling, and soft). This means that the participants felt solid and dark colored design as the comfort interior design.

Then, total comfort quantitative evaluation model was made using sound comfort evaluation model without or with considering the synergy effect and subjective design comfort score. Equation 5 and 6 show the total comfort evaluation model. Equations 5 and 6 shows the evaluation model without and with considering the synergy effect, respectively.

Total comfort = 7.46 + 0.62 × Interior design comfort

- 0.15 × Loudness - 20.69 × Partial roughness (under 10 Bark) (5)

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

Total comfort = 0.94 + Interior design comfort

× (2.39 - 0.042 × Loudness - 5.69 × Partial roughness (under 10 Bark) (6) 4. VERIFICATION OF TOTAL COMFORT EVALUATION MODEL

In this section, we carried out subjective evaluation test again using the same VR system for the verification of total comfort evaluation model. In the test, additional interior sound and design were prepared for the verification. As the presented interior sound, four samples were employed. Three of them (CS1, CS6, and CS9) sound samples were employed in the previous test. Another sample (CS11) was 15 dB reduction sound from 200 – 500 Hz by filtering CS6 to improve the comfort. As the interior design image, three designs were prepared for VR system as shown in Figure 8. One of them (ID1) was previously used design but the other two designs (ID5 and ID6) were used first in this verification test.

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW

(a) ID1. (b) ID5. (c) ID6. Figure 8: Presented interior design images in VR system for verification test.

Here, we estimated the total comfort score from the proposed total comfort model in two ways. The first model (Equation 5) does not consider synergy effect and another model includes the effect as shown in Equation 6. For estimation of the sound comfort, we calculated the loudness and rough- ness from the recorded sound pressure signal and the score was obtained using Equation 4. In case the design comfort score estimation, we did not have the calculation model as in the sound comfort estimation. Hence, we obtained the comfort score difference in color and the difference in shape from the first subjective test as shown in Table 1. Bold numbers are the obtained design comfort score in the first test and underlined numbers in the table shows the estimated design comfort score using the score difference in color and shape. Table 1: Sub jective and calculated design comfort score to each presente d interior design.

Color

Shape Dark Pastel Beige

Separated shape 4.08 （ ID1 ） 2.90 （ ID2 ） 3.62 （ ID6 ）

Combined shape 3.79 （ ID3 ） 2.61 （ ID5 ） 3.33 （ ID4 ）

Figure 9 shows the relationship between estimated total comfort score and the obtained subjective comfort score. Horizontal axis of Figure 9 (a) shows the estimated score without considering the synergy effect and the axis in Figure 9 (b) shows the score with considering the synergy effect. Ver- tical axes of both figures show the subjective total comfort score.

As shown in these figures, both relationship show high correlation, but the relationship considering the synergy effect in Figure 9 (b) was slightly higher than the relationship without considering the effect in Figure 9 (a).

Good

5

5

Good

4

4

Total comfort

Total comfort

Subjective

Subjective

3

3

2

2

R = 0.922 Bad

R = 0.934

1

1

Bad

1 2 3 4 5

1 2 3 4 5

Bad Estimated Total comfort Good

Bad Estimated Total comfort Good

(a) Without synergy effect (b) With synergy effect Figure 9: Relationship of total comfort score between subjective and objective scores. From these results, the influence of interior sound and design on total comfort in cabin is consid- ered to affect each other. In case the comfort of interior design is high, the influence of interior sound increases. And the total comfort is expected to increase largely by decreasing the loudness and rough- ness of the sound. 5. SUMMARY

In this study, we carried out a subjective evaluation test about the comfort in cabin using VR system to estimate the comfort quantitatively from the interior sound and design characteristic for the effec- tive improvement. From the subjective evaluating results, the total comfort was found to be expressed well using the sound and design comfort. In addition, the synergy effect was observed to express the total comfort better. Furthermore, the sound comfort was found to be increased by the soft and smooth sound. And the design comfort was high in case the interior design was dark and solid. From these results, the one side (sound/design) improvement is not enough to realize the comfort cabin, but the compatibility is found to be important for realizing the comfort cabin. 6. REFERENCES

1. Hashimoto, T. Improvement Sound Quality of Vehicle Noise. Journal of Automotive Engineer-

ing of Japan , 49(6) , 65–70 (1995). In Japanese 2. Sheng, G. Vehicle Noise, Vibration, and Sound Quality. SAE International (2012). 3. Fukuhara, C., Kamura, T. & Suetomi, T. Subjective evaluation of engine acceleration sound with

driving simulator. JSAE Review , 23(4), 435–441 (2002). 4. Yoshida, J., & Igata. T. Dependence of loudness evaluation by drivers on vehicle styling.

J.Acoust. Soc. Am ., 132 , 3866–3873 (2012). 5. Yamaguchi, M., Watanabe, Y., Maeda, O., Okubo, N. & Toi, T. Impression Change of EV

Driving Sound in Consideration of Audio-visual Information and Driving Intention. Transac- tions of JSAE, 45(5) , 859–864 (2014). In Japanese 6. Yoshioka, E. Itou, S. & Yoshida, J. Influence of steering vibration on vehicle speed recognition

and comfortableness in cabin, Proceedings of Inter-Noise 2021, pp.1954–1962 (2021). 7. Fastl, H. & Zwicker, E. Psychoacoustics: Facts and Models, Springer , (2006). 8. Hoshino, H. Evaluation of passenger car interior noise based on an auditory processing model in

consideration of masking effect. Journal of Acoustical Society of Japan , 56(1) , 3-12 (1989). In Japanese

i, orn inter.noise 21-24 AUGUST SCOTTISH EVENT CAMPUS ? O? ? GLASGOW