Intercultural Differences in the Perception of HVAC Sound Quality in Car Cabins: From Conventional to Electric Vehicles

: In electric-powered cars, the production of which is increasing, the HVAC system is responsible for most of the noise inside the car’s cabin, causing significant discomfort for passengers. Moreover, the noise produced by the HVAC affects the perceptible sound inside the car cabin, significantly impacting the perceived quality of the vehicle. It is thus essential to investigate and quan-tify people’s preferences concerning HVAC noise. Our previous research revealed differences in the HVAC noise between hybrid electric (HEV) and internal combustion engine (ICEV) vehicles. A sub-sequent factor analysis revealed that the adjectives used to describe the sounds can be grouped into two main dimensions: Aesthetic and Loudness. The present paper highlights the results of a listening test that aimed to identify possible differences in the perception of HVACs’ sound quality between Italian and Japanese subject groups, for ICEV and HEV, in different functioning conditions. Results revealed that the most remarkable difference emerges at high air flow rates, where the Japanese group perceived the quality of sound and annoyance, respectively, to be significantly lower and significantly higher than the Italian group.


Introduction
Car users have become increasingly demanding [1,2], asking for higher sound quality levels inside vehicles' cabins. Sound quality consists of a perceptual reaction to the sound of a product that reflects the listener's reaction to the acceptability of that sound; the more the sound is considered to be acceptable, the greater the sound quality [3,4]. However, having low-level sound does not necessarily mean having high sound quality [5]. To investigate the complexity of sound quality perception inside vehicles, several authors have developed models and metrics that consider the different aspects of perceptual dimensions [6][7][8][9][10].
Heating, Ventilation, and Air Conditioning (HVAC) systems are fundamental to set and maintain the microclimatic conditions inside a car's cabin or its portions, providing optimal thermal comfort for occupants. Nevertheless, these systems represent one of the most annoying noise sources for drivers and passengers. This is particularly true in idling conditions, at low speed of Internal Combustion Engine Vehicles (ICEV) and, at low engine rpms.
HVAC noise is produced by the fan rotation and airflow interactions passing through the different plant components (e.g., outlets, thermal exchangers, and ducts) [11]; consequently, this noise is generated not far from the receiver's positions of drivers and passengers. This has focused the attention of the automotive industry to propose different noise reduction design measures on moving all or part of the HVAC system into the engine compartment or using passive acoustics absorbers on the propagation path [12] to improve the acoustic and thermal comfort [13,14].
As a result of the increasing number of hybrid electric vehicles (HEVs) [15], the character of noise inside cabins has significantly changed [16]. Graham-Rowe et al. [17] found that the lack of engine noise when using electric vehicles leads many drivers to experience pleasant driving. However, this is not the only reaction. Because customers accept or at times even expect and wish for operational feedback in certain situations while driving electric, some drivers consider the acoustic feedback of car inadequate [18,19]. For Krishna this can be associated with the culture of cars [20]. Previous research regarding nationality or culture found that sound preferences of door sounds were similar between German and Japanese participants [21], but differences were revealed between Americans and West Europeans in an evaluation of the vehicle interior sound during driving [22].
Due to the lack of the ICEV engine noise, the background noise in HEVs has been dramatically reduced by up to 9 dB [16]. As a result, the role of the HVAC noise on the perceived sound quality in cars' cabins is becoming more crucial.
This paper presents the results of an intermediate step of the "Collaborative Research Project on HVAC sound quality inside cars cabins" between the Department of Architecture and Industrial Design of the Università degli Studi della Campania "Luigi Vanvitelli" and the Department of Communication Design Science, Faculty of Design, Kyushu University. In this phase of the project, some of the previously collected binaural recordings were used to investigate, in two parallel listening tests in Italy and Japan [23,24], the possible differences in sound quality perceptions between two types of engines in Italian and Japanese populations.

Audio Material
The recordings were of two groups of vehicles, ICEVs and HEVs. In total, 20 units were measured-in Japan, 8 HEVs and 2 ICEVs, whereas in Italy 2 HEVs and 8 ICEVs. In all cases, HVAC noise was measured at the front-left seat inside each vehicle, i.e., the driver position in Italy, the passenger position in Japan. The background noise inside the vehicles was less than 35 dB in the equivalent continuous A-weighted sound pressure level (LAeq). Measurements were carried out with a 4-channel measuring system. This consisted of a SQobold, a binaural headset (Head Acoustics/BHS II) positioned on a Mk2 Cortex manikin in Italy; and a 2-channel head and torso simulator (HATS Type 4100, Brüel & Kjaer), in Japan. The playback sound pressure level was calibrated using, respectively, a binaural Mk2 Cortex manikin, and a head and torso simulator HATS. The position of the right microphone of the HATS/BHS II was approximately 0.80 m from the closest central outlet and about 0.70 m above the seat. During the measurements, all the windows and doors were kept closed, and all the air outlet valves were fully opened and slightly oriented to the floor to prevent noise induced by airflow on the microphones. Recordings were made compatible using a comparative measurement carried out at the Kyushu University and by developing equalizer filters. Each recording lasted 30 s and considered different Function Modalities (FMs), Air Outlets (AOs), and Air Flow Rates (AFRs) [4]. Sections having a length of 5 s were analyzed by the software (Artemis Suite, Head Acoustics) to extract representative audio samples. To avoid a click noise, 60 ms fade-in and 80 ms fade-out were processed for each stimulus. The excerpts were then grouped into two Functional Modalities (Ventilation and Air-Conditioning) and Engine Types (ICEV and HEV) as shown below: Concerning the Air Outlets, only the frontal one was considered because it represents the most common and frequently used. Furthermore, concerning the Air Flow Rates, four AFRs were considered, when the ventilation power was at 25%, 50%, 75% and 100%.
Finally, the collected samples were compared, and those within one standard deviation on the A-weighted sound pressure level were chosen. In total, 48 HVAC samples were used in the experiment-2 FM × 4 AFR for each test vehicle, i.e., 3 ICEVs + 3 HEVs.

Subjective Questionnaire
Participants were asked to rate their impression of each stimulus on a 7-point semantic differential scale of 14 adjective pairs, which explored different aspects of sound: Quality, Pleasantness, Power, Spatio-temporal, and Spectral structure [25][26][27][28]. The scales were originally collected in English, then the translated versions, in Japanese and Italian, were used in the experiments (Table 1).

Japanese
English

Participants
Forty volunteers participated in two parallel experiments. Twenty of them (10 males; 10 females) (Mage = 22.1; SDage = 1.07) were enrolled in the experiment in the Japanese Subject Group, and the other 20 (13 males; 7 females) (Mage = 30.6; SDage = 5.1) were enrolled in the experiment in the Italian Subject Group. All participated voluntarily, without receiving any type of compensation.

Method
Two parallel listening experiments were conducted to investigate the perception of the HVAC noise in car cabins. The experiments were performed in soundproof laboratories at Kyushu University and Università degli Studi della Campania "Luigi Vanvitelli". All 48 stimuli were presented to participants via an audio-interface (RME/babyface) and headphones (Sennheiser/HD650) in Japan, and via laptop and headphones (Sennheiser/HD200 PRO) in Italy. After participants received information about the study and provided informed consent, the procedure started. Participants could listen to the stimuli as many times as they wanted before proceeding to the next one. Three adjective couples most pertaining to Aesthetic and Loudness dimensions [12,16] were chosen from the experimental sessions. They are High Quality-Low Quality, Not Annoying-Annoying and Unpleasant-Pleasant for the Aesthetic dimension; and Loud-Soft, Noisy-Quiet, and Powerful-Weak, for the Loudness dimension.

Results
To investigate possible differences between the two Subject Groups in the Aesthetic ( Figure 1) and Loudness (Figure 2) dimensions, for each of the adjective couples, two repeated measures ANOVAs were conducted separately for the Ventilation and Air Conditioning Functional Modalities. The ANOVAs treated Engine Type and Air Flow Rate as within-subjects factors and the Subject Group as a between-subjects factor.
Considering the adjectives of the Aesthetic dimension, the analyses of the High Quality-Low Quality (    Marginal means of the previous three adjective couples describing the Aesthetic dimension as a function of the Engine Type and Air Flow Rate are shown in Figure 1. Different graphs are shown for the Ventilation and Air Conditioning Functional Modality.   (Table 5) and Noisy-Quiet (  Marginal means of the previous three adjective couples describing the Loudness dimension as a function of the Engine Type and Air Flow Rate are shown in Figure 1. Different graphs are shown for the Ventilation and Air Conditioning Functional Modality.

Discussion
In line with the literature [29,30], previous research findings showed that [23,24] Aesthetics and Loudness represent the two main dimensions describing the perception of HVAC noise inside car cabins. In this research, the three most representative adjective couples describing each dimension were considered.
For the Aesthetic dimension, the listening test results for the adjective couples of High Quality-Low Quality, Not Annoying-Annoying, and Unpleasant-Pleasant highlighted that the most relevant difference between the two Subject Groups occurred for High Quality-Low Quality. More specifically, the Japanese considered the quality of sounds to be significantly lower than Italians in both ICEVs and HEVs and whether the AC was turned on or off, but only at the highest air flow rates. The Not Annoying-Annoying item was found to be affected mainly by the airflow rates and the engine type, with ICEVs considered more annoying than HEVs, especially at low AFRs. Moreover, both the airflow rate and the type of engine showed interaction effects with the subject group in the Ventilation condition, with Japanese considering ICEVs more annoying. This interaction effect disappeared when the air conditioning was turned on. Furthermore, the results for "Unpleasant-Pleasant" also showed the main effects of the air flow rates and the engine type in influencing the evaluations. The higher the air flow rate, the more unpleasant the judgements. Both groups considered HEVs more pleasant.
For the Loudness dimension, Loud-Soft, Noisy-Quiet, and Powerful-Weak adjective couples were considered. For Loud-Soft, the air flow rates and type of engine affected the evaluations, whether with or without air conditioning on. In general, HEVs were considered softer then ICEVs but both were evaluated as loud as air flow rate increased. Regarding Noisy-Quiet, similar to the previous adjective couples, HEVs were considered quieter then ICEVs and both were evaluated as noisy as the air flow rate increased. Finally, considering the Powerful-Weak adjective couple, results showed the main effects of air flow rates and engine type in influencing the evaluations, with ICEVs considered significantly more powerful than HEVs by Japanese, especially at high air flow rates.
Although the Air Flow Rates of 75% and 100% were responsible for the general worsening in the Aesthetic and Loudness items for both groups, thus minimizing the positive effects of passing from ICEV to HEV, the findings suggests that Japanese market designers should control the effects of the high Air Flow Rates as much as possible, because they dramatically affect the perception of low quality and annoyance of the HVAC sound, more so than for Italians.
Although this research investigated the perception of HVACs' sound quality in several auditory scenarios-four Air Flow Rate levels, two Engine Type levels, and two Functional Modalities-per each Subject Group, their combination represents only some of the auditory scenarios that drivers and passengers may experience every day. In future, more realistic and complex operative conditions, which also consider the interactions among the HVAC systems, the cars operational conditions (i.e., driving speeds, urban vs. extraurban), and human habits and preferences (i.e., use of infotainment systems) should be investigated.