FabSound: Audio-Tactile and Affective Fabric Experiences Through Mid-air Haptics

The sound produced when touching fabrics, like a blanket, often provides information regarding the fabric’s texture properties (e.g., its roughness). Fabric roughness is one of the most important aspects of assessing fabric tactile properties. Prior research has demonstrated that touch-related sounds can alter the perception of textures. However, understanding touch-related sound of digital fabric textures, and how they could convey affective responses remain a challenge. In this study, we mapped digital fabric textures using mid-air haptics stimuli and examined how auditory manipulation influences people’s roughness perception. Through qualitative interviews, participants detailed that while rubbing sounds smoothen fabric texture perception, pure tone sounds of 450Hz and 900Hz accent roughness perception. The rubbing sound of fabric evoked associations with soft-materials and led to more calming experiences. In addition, we discussed how haptic interaction can be extended to multisensory modes, revealing a new perspective of mapping multisensory experiences for digital fabrics.


INTRODUCTION
When we interact with a fabric (e.g., when selecting garments in the morning or choosing a blanket in a shop), we perceive not only its texture but also the sound it produced while touching it.This combination of sensory cues is often referred to as audiotactile perception [23], and help people to make more informed assessments about the roughness of textured materials.However, studies have shown that auditory cues presented alone can be used to enhance the user's tactile experiences of textured surfaces and their resulting behaviours [19,21,22,57].For example, Guest et al. [23] showed that auditory cues alone can provide sufcient information to determine the size and material of objects.Jousmäki and Hari [27] reported that altering the sound frequency of a user's hands rubbing together can make their hands feel smoother or rougher.These studies suggest audio feedback is a powerful tool to improve user touch perception when interacting with tactile patterns, enabling more robust texture estimation [8,32].However, the vast majority of audio-tactile research has focused on physical, rather than digital tactile stimulation.
In recent years, contactless haptic technology, such as ultrasoundbased mid-air haptics, has rapidly developed, allowing people to experience tactile sensations without any physical touch.Mid-air haptics technology is modulating ultrasound to create focal points on the human skin, producing tactile sensations without physical contact [7,18,38].
New developments in this technology have achieved the rendering of mid-air textured surfaces that can be felt, involving complex haptic attributes such as roughness [4,18,36], stifness [33], softness [33], and viscosity [26].Previous studies have attempted to map mid-air haptics textures with the texture of physical materials [4].However the texture of fabric has remained under-explored despite the potential of its tactile properties (e.g., roughness, stifness, compression, and temperature [28,40,57]) to elicit a wide range of afective responses [57].Therefore, it still remains challenging to deliver convincing fne detailed fabric textures to users through mid-air haptics technologies.
In this paper, we aimed to understand whether this contactless mid-air haptics technology could be used to convey the fne details of nine fabric textures, and if users could efciently map the similarity between the physical fabric samples and the mid-air haptics fabric textures.To address this questions, we conducted two studies.In the frst study, using only tactile stimuli, we assessed how closely 9 physical cotton fabrics and their mid-air haptics counterparts are perceived.Results showed that users found it hard to match physical and digital textures, however, they were able to fnd diferences in roughness between the nine mid-air haptics fabric textures.In the second study, we combined one mid-air haptics texture (i.e., velvet) with diferent touch-related rubbing sounds and sine waveform pure-tone sounds (i.e., 100Hz, 450Hz, and 900Hz) to explore if those auditory cues can alter the participants' tactile sensations of the mid-air haptics texture stimuli.
Using a mixed-methods approach, combining semi-structured interviews and questionnaire analysis, our work -FabSound, revealed that by changing only the auditory cues, participants explained they were amazed by how diferent the same digital fabric felt in terms of roughness and the texture that they perceived, even though it was always the same texture stimulus.Our results demonstrated that users' perception of roughness can be altered using both touch-related sounds and pure-tone sounds for mid-air haptics texture experiences.Moreover, through integrating sounds with fabric textures perception in mid-air, users can have a richer afective response including association to fabrics and past experiences, resulting in an enriched experiences to the users.Through this audio-tactile integration, online E-commerce could extend to multisensory experiences, resulting in a more engaging experience for users.

RELATED WORK
Our perception of fabric is multisensory [11,23,57].Whenever we choose a garment for work or a blanket in a store, we evaluate its textures simultaneously using several senses.Through the lenses of textile research, psychology, and auditory research, we present recent studies and discussions of the infuences of sound on both the texture and roughness perception of fabrics below.

Roughness of fabrics and textures
Roughness is one of the most important and efective factors in predicting the tactile properties of fabrics [35].In terms of surface roughness, there are two major categories: subjective and objective.It has been shown that the way fabric is handled by people is largely infuenced by the fabric's physical properties and how comfortable it feels subjectively.The subjective evaluation process is often carried out by a method known as textile handling [5,16,28,42].During textile handling, the fabric is touched and bent with the fngers, then slightly stretched with the hand [42,57].Particular tactile properties, such as roughness, are assessed during the subjective evaluation process [5,16].The subjective nature of such evaluation methods is also supported by psychological studies of tactile perception.Researchers have highlighted the richness of tactile and afective responses and discussed roughness as one of the most prominent tactile dimensions of surface textures [40,57].It is also important to objectively measure the tactile properties of textiles from a textile material research perspective.To measure fabric tactile properties (such as roughness, compression, bending, and so on), several methods were proposed, including Kawabata Evaluation System and Fabric Touch Tester [25,28,57].Nevertheless, both subjective and objective measurements emphasize fabric roughness as an important tactile characteristic.
However, it is difcult to render the subjective and objective fabric properties in the digital space.Current surface texture haptics research often uses virtual textures and surface features to simulate material roughness, provide informative force feedback, and augment spatial elements [1].For example, Bau et al. [2] shown that a time-varying periodic signal can be used to synthesize virtual textures by actuating a friction display.The signal can evoke natural sensations of roughness such as touching wood, bumpy leather, paper, a painted wall, or rubber.Although attempts have been made to recreate fabric textures in digital space, there is still a lack of sophisticated tactile feedback displayed through touch surfaces results in a decrease on user experience quality and even task performance.While tactile rendering techniques have advanced signifcantly over the past few decades, they still do not achieve the level of realism required to simulate a wide range of textures.It is likely caused by haptic technologies' limitations [1], which could be augmented through audio-tactile experiences, which we will discuss next.

Audio-tactile roughness perception
When congruent information is presented to both sensory modalities simultaneously, tactile cues often dominate auditory cues in determining texture perception [29].The tendency to focus attention on tactile information may be due to low-level sound cues in many everyday situations often being masked by the general background noise [29].However, psychologists have noted that noise bursts or repeated tones can afect the tactile perception of roughness [13].As a result, auditory and tactile information may be integrated multisensorially in roughness perception [11,49].Recently, Murari et al. rated six major pure tones and six minor 30-second pure tone fragments based on seven sensory factors [37].Participants tended to match higher roughness values to minor pure tones.Similar results are also demonstrated in Wallmark et al. 's work [53], where preschoolers crossmodally associated tactile roughness to low-pitch sounds, with no associations to visual experience.Moreover, Hamilton-Fletcher et al. [24] compared soundtouch correspondences in sighted and blind adults to determine whether visual experience would infuence the strength and direction of sound-touch crossmodal associations.Their research suggests early and late blind individuals tend to associate low frequency pure tone sounds with rough textures, and high frequency pure tone sounds with soft and smooth textures.In sighted individuals, similar associations were also observed by Eitan and Timmers [15] who studied verbal associations between sounds and words like roughness and smoothness.
Moreover, Jousmäki and Hari [27] conducted a study in which participants rubbed their hands together and simultaneously heard the increased/decreased sound that was made by the rubbing through headphones.Participants were asked to rate the feeling of the skin on their palms on a scale of smooth to rough.As a result of increasing the sound frequencies, participants tended to rate their hands as smoother [27].This change in the perception of roughness provides intriguing evidence regarding auditory feedback's infuence on tactile perception.This was further studied by Guest et al. [23], who found that increasing the frequency of the rubbing sound of an abrasive texture can result in an increase in roughness perception.According to their study, the auditory cues signifcantly shape texture surface perception of tactility and emphasize the multisensory nature of roughness perceptions.
The research presented so far supports that both pure tone sounds (i.e.specifc frequencies) [37,53] and touch-related sounds (i.e.diferent rubbing sounds) [23,27] can afect the perceptions of fabric roughness.However, the question of how audio-tactile cues infuence our perception of fabric roughness in a digital space remains unexplored.

Afective experiences of fabrics
Recent studies have proven that the fabric experiences in our everyday life are not only sensorial, but also afective in nature [57].Researchers have found that touching an object elicits an afective response (e.g.pleasant or unpleasant reactions), and the intensity of activation is greater in the emotional regions than those in the sensory regions [47].Further to this, a recent study by Xue et al. demonstrated that the tactile experiences of fabrics can elicit a wide range of afective experiences including association, valence (unpleasant or pleasant), arousal (calming or exciting) and preferences.They emphasize the importance of engaging diferent sensory cues to design digital fabric experiences [57].The same concept was also highlighted in consumer research.Li et al. [31] illustrated in their study that consumers' sensory perceptions and emotions can have a profound impact on their purchasing decisions.Recent studies in mid-air haptics tactile experiences also suggest that mid-air haptics stimuli have a great potential to communicate emotional responses to users through touch [39].
Despite the rapid development in digital immersive interaction design, there is still a lack of understanding of users' multisensory experiences (such as audio-tactile roughness perception) and afective responses.Here, we addressed these gaps by using Beattie et al. 's [4] methods to map nine fabrics' textures into mid-air haptics.Using mid-air haptics fabric textures, we examined whether these textures could convey the same tactile property of roughness and evoke emotions such as pleasantness or calmness/excitement.

Mid-air haptics texture perception
Recent developments of mid-air haptics technology have attempted to render textural properties like "roughness" [4,17,18].For instance, Freeman et al. [18] described the method of rendering surface textures (e.g.metal gratings) using macro-scale texture rendering.They defned the surface as the geometric features [18].Beattie et al. [4] outlined a method to recreate textural features in images.Specifcally, images were analyzed to defne degrees of roughness, which were then mapped to haptic features [4].Their study explored how manipulating particle color, size, distribution, density, bumpiness, and arrangement afect the accuracy of haptic sensations.Their results suggest that the amplitude modulation are highly related to three visual texture parameters: particle density, bumpiness, and arrangement [34,56].
However, fabric textural properties can be difcult to render convincingly through mid-air haptics, as the fne features that typically make up the texture of fabrics are much smaller than the focal point of ultrasound [45].As a potential solution to these limitations, auditory cues are introduced to enhance the tactile sensation when users explore the textured surface (e.g.fabric surface) in mid-air haptics texture rendering.
Although prior work in HCI has extensively studied the efect of audio-tactile feedback on the user experience in physical interactions [19,21,22,57], there have been limited studies exploring audio-tactile feedback in mid-air haptics.Studies in haptic technology begin to outline how adding auditory stimulation to mid-air haptics can modulate our afective and tactile responses to digital textures [19,20,34,38,41,52].In a recently published work, Montano-Murillo et al. [34] explored how audio and mid-air haptics could infuence people's perceptions of texture, body sensation, and motor behavior.The researchers found that audio and haptics can be used to infuence texture attributes, hand sensations, and hand speed when exploring mid-air haptic textures.However, their study focused primarily on surface textures such as wood and metal instead of fabrics, suggesting that multisensory experiences can assist in conveying tactile information more efectively.These fndings highlight that audio-tactile feedback in mid-air haptics can be used to create a more immersive multisensory experience, as the user can both feel and hear the haptics feedback.It is concluded that by combining mid-air haptics with audio feedback, users may experience tactile and immersive experiences that are enhanced by multisensory experiences.Hence, in this study, we examined the infuence of both pure tones and touch-related sounds (such as fabric rubbing sounds) on mid-air haptics fabric textures.Audiotactile experiments were also explored to determine if they would enhance user afective responses to texture stimuli.

USER STUDY 1
While some studies have explored tactile experiences through midair haptics stimuli, none have examined how diferent fabric textures can be mapped through mid-air haptics.As part of our frst study, we examined whether users' tactile experiences with fabric textures in mid-air haptics are diferent from those with physical fabric textures.In our Study 1, we aim to a) determine if the mapped mid-air texture stimuli resemble fabric textures in real life; b) investigate into how participants perceive the roughness of the mapped mid-air haptics textures and c) investigate whether mid-air haptics texture stimuli can convey emotions of valence (unpleasant or pleasant), arousal (calming or exciting).Earlier research has found that tactile roughness is equally perceptible with the right and left hands when tactile roughness is explored laterally [30].Therefore, in this study, participants was asked to actively explored both mid-air haptic fabric textures and physical fabric textures simultaneously by both hands.The study was approved by the Ethics Committee of the local university (the approval number was anonymized).A written informed consent was obtained from all participants before they took part in the study.

Study design and method
In this study, we frst selected 9 fabric samples based on a recent study conducted by Xue et al. [57].The fabric samples were then mapped into mid-air haptics stimuli using the method proposed by Beattie et al. [4].Following this, participants were asked to compare the physical fabric samples and the corresponding mid-air haptics texture to rate the similarity, roughness, valence and arousal of the stimulus.Here, we present our methods in details.
3.1.1Fabric selections.The mid-air haptics texture was mapped to haptics textures using nine diferent cotton fabrics (F1 -F9) [57].They are F1-Heavy Drill, F2-Flannel, F3-Buckram, F4-Organza, F5-Velvet, F6-Voile, F7-Calico, F8-Muslin, and F9-Twill.Details of each fabric is shown in Figure 2 (a.).Each physical fabric sample was made from 100% cotton.Cotton was selected due to the fabrics ubiquity and various tactile textures [57], meaning it ofers a wide range of stimuli suitable for investigating roughness.Moreover, according to a recent study conducted by Xue et al. [57], these nine cotton fabrics can elicit a wide range of tactile and afective responses.These properties of the nine samples make them excellent choices to explore whether the mid-air haptics texture could convey a similar feeling to users.

3.1.2
Haptics texture rendering of the fabrics.We delivered mid-air haptics textures to participants with the Ultraleap STRATOS Explore Development Kit (16 × 16 transducer array boards).We tracked participants' hand movements throughout the study using an Leap Motion Controller LM-010.Mid-air haptics textures were rendered using an algorithm that maps visual attributes of textures into mid-air haptics patterns proposed by Beattie et al. [4].As part of the method, visual cues like spatial distribution of surface elements are identifed from any two-dimensional graphical image, then haptic attributes are used to replicate those cues to create texture representations.Open-source implementation of the algorithm can be found at [3].
Based on this method, we frstly connected a Digital Microscope Camera with magnifcation range from 40X to maximum of 1000X to a Mac book Pro laptop to capture the texture image of each of the nine fabrics.Details for each image is shown in Figure 2 (a).In order to obtain mid-air haptics textures, we replicated the method used by Beattie et al. [4] and Montano-Murillo et al. [34].This created digital counterparts of each of the nine fabrics (F1 -F9), referred to as H1 -H9 (stimulus).A summary of this mapping process is shown in Figure 2 (b).More specifcally, we generated each of the haptics textures by the following steps: 1. Creating Displacement and Normal Maps: displacement and normal maps are created for each fabric texture using a specialized online tool.Specifc parameters, including contrast (-0.8), strength (3.5), level (0.8), displacement (0.4), and the Sobel flter, are adjusted to generate both the normal map and its corresponding displacement map [34].To enhance the quality, each displacement map undergoes a cleaning and regeneration process in Adobe Photoshop 2023 (version 24.7.0), resulting in a clearer and more refned displacement map. 2. Micro-Roughness Mapping: We used the gradient of the power spectral density function to compute microscale roughness from the displacement map, as described by Beattie et al. [4].We used this information to determine the frequency at which small texture changes occur (high frequency), which produces smoother textures.3. Macro-roughness Mapping: We then used displacement map values to calculate macro-scale roughness (black and white forms).4. Creating Haptic Feedback: By using the look-up table for texture roughness based on the pre-computed values from the previous steps, we converted micro-roughness to Ultraleap rotation speed and waveform sampling parameters.Over texture points, we used the middle fnger position to calculate the rotation speed and focal point intensity of the haptics device.A circle stimulus of 2 cm radius was used for each texture point, set to maximum intensity when hovering a white pixel and to minimum intensity when hovering a black pixel.
We named the mapped mid-air haptics fabric texture stimulus as [H + number of 1-9 + digital + fabric sample name] in order to simplify the name of each mid-air haptics stimulus.As an example, F1-Heavy Drill's mid-air haptics texture mapping is designated H1-Digital Heavy Drill, while F2-Flannel is designated H2-Digital Flannel, and so on.

Study setup and procedure
The mid-air haptics device was placed inside a black box with a height of 26cm.A 20 × 20 cm square was cut out of the top so participants can assess the mid-air haptics texture when they rest their wrist on top of the black box.For the physical fabrics, 9 samples were cut into 20 × 20 cm squares and mounted on white A3 card (as show in Figure 3 a).Each of the fabric was insert into the bottom of a diferent black box, which had a cut-out at the front and back.The researcher would add fabrics into this box from the back and users would feel the fabrics from the front, where a curtain obscured their visual perception of the fabrics.As users would feel the digital and physical fabrics simultaneously, the fabric sample box was raised to the same height as the mid-air haptics stimulus.
In the study, participants were required to sit on an adjustable chair in front of a computer screen with a mouse and keyboard.The frst black box with the mid-air haptics device was placed on the right side of the table, where participants could access by using their right hand.The second black box where participants could assess the physical fabric samples using their left hand was placed on the left side of the table.A pair of 3M Noise Cancelling headphones was provided to the participants to reduce any auditory infuences throughout the study.The whole setup for the study is shown in Figure 3 (b.).
Following introduction of the study procedure to each participant, we provided them with a training stimulus that would help them become familiar with the testing procedures and stimuli.After familiarisation, participants explored a fabric and a mid-air haptics stimulus simultaneously.In total, there were 81 trials comparing each of the nine digital and physical fabrics to one another.Each exploration lasted 20 seconds.The participants were asked to fll out four questions after each exploration.The four Likert-scale questionnaires were: a) how similar the fabric was to a mid-air haptics stimulus from 1-not similar at all to 7-very similar; b) how rough do they feel about the mid-air haptics stimulus from 1-very smooth to 7-very rough, c) how pleasant they feel about the mid-air haptics stimulus from 1-very unpleasant to 7-very pleasant, and d) how activate they feel about the stimulus from 1-very calming to 9-very exciting.The whole process of the study procedure is show as in Figure 3 (c.).The whole study lasted no more than 1 hour.

Participants
During the study, 37 participants volunteered (18 females, 19 males, mean age ± SD: 28.7 ± 7.3 years, range = 20-51 years), none of whom reported any impairments afecting their perception of mid-air haptics stimuli (e.g.neuropathy or vascular problems).All participants were recruited from the university's participant pool and were compensated for their time with a gift voucher of approximately $12.5.In the study, 21 participants had no prior experience with midair haptics devices, 6 participants had limited experiences, and 10 participants had extensive experience with mid-air haptics devices.

Analysis and results
All data collected were analysed using IBM SPSS Statistics analysis software (version: 28.0.0.0) and visualisations were created using R Studio 2021 [44] and the ggplot2 [55] and gghalves packages [50] .For each digital fabric, we conducted a repeated measures ANOVA comparing similarity ratings between each of the physical fabrics.This process was also repeated for the other dependent variables (roughness, valence and arousal).

Study 1 summary and discussion
As a result of our similarity test between the 9 physical fabric samples and the 9 mid-air haptics fabric textures, fabric sample F3 (Buckram) consistently exhibits the lowest similarity ratings across diferent stimuli compared to other fabric samples.Participants rated the fabric sample -F3 (Buckram) is the least similar material to the mid-air haptics texture stimuli during the study.The main reason for this could be a lack of compression in the mid-air haptics texture stimuli.For example, one participants stated after the test: "When I touch the mid-air haptics and the fabric together, I have difculty comparing them because there are no compression on my hand with the stimulation.My hand felt no pressure when I pressed down (on the stimulus), and when I pressed down on the fabric, I could feel a force pushing my hand against my palm" [P009].Particularly when it comes to Buckram (material F3, which is often used for women's summer hats) has a higher degree of roughness and strength compared to other materials.There is no clear mapping between F2 (Flannel), F4 (Organza), F5 (Velvet), F6 (Voile), F7 (Calico), F8 (Muslin), and F9 (Twill) to the mid-air haptics texture stimuli H1-H9, and they cannot determine the exact relationship between each mid-air haptics texture stimulus and the physical fabric textures.In short, participants fnd all 9 mid-air haptics stimuli are similar to the physical fabric texture of F2 (Flannel), F4 (Organza), F5 (Velvet), F6 (Voile), F7 (Calico), F8 (Muslin), and F9 (Twill).
The results of the roughness ratings shown a signifcant diference in perceived roughness among nine mid-air haptics stimuli.According to our analysis, participants identifed the least rough haptic textures stimuli, H6-Digital Voile and H7-Digital Calico.Additionally, participants perceived H2-Digital Flannel as having a rougher texture than H6-Digital Voile and H7-Digital Calico.
Participants, however, perceived all nine mid-air haptics textures as having a smoother surface and there was little variation between textures.Additionally, participants found that the mid-air haptics fabric texture stimuli did not difer in terms of their emotional Moreover, in contrast to prior literature on emotional responses responses.In general, participants found that all mid-air haptics to physical fabric texture by Xue et al. [57], there was a lack of texture stimuli tended to be more similar to the smooth surface emotional responses to digital mid-air haptics fabric textures.This physical fabric samples.It is likely due to the contactless, ultra-led us to explore whether auditory cues could afect the roughness sound sensation that is more subtle than any physical stimulus that perception of the mid-air haptics fabric texture and whether the provides force feedback.
audio-tactile experiences of the mid-air haptics texture could alter the mid-air haptic fabric texture perception and elicit a wider range of emotional responses.

USER STUDY 2
The purpose of this study was to evaluate whether the audio-tactile experiences could afect the perception of touch when users interact with mid-air haptics stimuli.As participants were perceiving all the mid-air haptics texture as a soft surface in Study 1 (see section 3.5), we selected one of the soft texture surface texture (velvet) to be combined with auditory feedback to test a) whether participants could perceive roughness diferences; and b) whether audio-tactile experiences could infuence their afective responses.As texture roughness is perceived equally by both hands when explored actively [30], we asked participants to explore mid-air haptic textures with their right hands in this study.The study was approved by the Ethics Committee of the local university (ethics approval number: UCLIC_2021_014_ObristPE).A written informed consent was obtained from all participants before they took part in the study.

Study design and method
To understand how auditory cues afect tactile perception of mid-air haptic fabrics.Our study examined whether touch-related sounds could infuence perceived roughness perception of mid-air haptic texture stimuli via a selection of recorded rubbing sounds.Furthermore, we examined how frequency and volume adjustments afected perceived roughness for a diverse audience, including designers, practitioners, and tech developers.Individuals can fne-tune perceived roughness by selecting the frequency or volume manipulation method strategically.When limitations arise that restrict access to only one audio manipulation method, this fexibility is particularly important.With this study, designers, practitioners and developers have both frequency and volume adjustments available to allow them to navigate constraints creatively, despite potential limitations.In addition, we investigated the efect of pure tone sounds on roughness perception as suggested by Hamilton-Fletcher et al. [24] and Eitan and Timmers [15] by selecting low, medium, and high frequencies (100Hz, 450Hz, and 900Hz).
As part of this study, we frst recorded the rubbing sound of the selected fabric.Then we recorded a sine wave form pure tone at 100Hz, 450Hz, and 900Hz to combine with the mid-air haptics texture stimuli.We then used a mixed method approach combining semi-structured interviews and questionnaire analysis to obtain a more comprehensive understanding of the users' audio-tactile experiences.Each method is described in detail in this section.tone sounds at three diferent frequencies of 100Hz, 450Hz, and 900Hz using an online sound generator [46].
For this study, the recorded audio was paired with the mid-air haptics texture stimulus obtained from Study 1 (H5-Digital Velvet) to create audio-tactile stimuli.For comparison with audio-tactile mid-air haptics texture stimuli, one stimulus has no audio input.In total, nine stimuli was used in this study.We name each of the stimulus in the format of A + number 1-9 -audio manipulation methods.For example, the midair haptics texture combined with recorded fabric rubbing sound with reduced frequency is named as A3-Frequency of -12dB, and the midair haptics texture combined with 100Hz pure tone frequencies are named as A7-100Hz Sound, and so on.Detail of each stimulus is listed in Table 2.

Mixed methods approach.
To investigate the audio-tactile texture experiences of the users, we combined quantitative (questionnaire) and qualitative (semi-structured interview) methods.Participants was asked to answer three questionnaires after they explored each audio-tactile mid-air haptics texture stimulus before the interview.Three questionnaires are: a) how rough do they feel about the mid-air haptics stimulus from 1-very smooth to 100-very rough; b) how pleasant they feel about the mid-air haptics stimulus from 1-very to 100-very unpleasant; c) how activate they feel about the stimulus from 1-very calming to 9-very exciting.We adapted the questionnaires scales from 1 to 7 range to 1 to 100 to avoid grouping responses around the tick mark [12].Then, the participant was interviewed after they fnished with the questionnaire.We guided each participants with a set of guiding questions including: "Please think back to the moment when you frst touch the mid-air tactile stimulus.In that moment, can you please tell me, where do you feel the stimulus on your hand?", "How does the stimulus feel on your palm/fnger (position on hand)?", "Think back about how the mid-air haptics stimulus felt on your hand, palm, and fngers.Can you please describe how rough or smooth it felt?","When you frst touch the mid-air stimulus, what does come to your mind, if at all?", and "Does this stimulus remind you of anything, or not?"This set of guiding questions was further extended based on the participants' responses, allowing an open conversation between the researcher and participants.

Study setup and procedure
The mid-air haptics was placed within a black box with a square cut out on the top.In the study, participants were required to sit on an adjustable chair in front of a computer screen with a mouse Figure 8: Overview of the study process: for each audio-tactile stimulus (A1 -A9), participants went through a familiarization phase, explored the audio-tactile stimuli, followed by questionnaires and fnally took part in a semi-structured interview.This process are repeated 9 times.and keyboard.On the right side of the table was a black box with a mid-air haptic device that participants could access with their right hands.The full setup of the study is show in Figure 8.
During the study, audio feedback was delivered using a studio headphone (noise-cancelling Sennheiser 400s headphones).Following introduction of the study procedure to each participant, we provided them with a training stimulus that would help them become familiar with the testing procedures and stimuli.After familiarisation, participants explored a mid-air haptics stimulus while a sound was simultaneously feedback to them for each exploration.Each exploration lasted no more than 30 seconds.The participants were asked to fll out a questionnaire regarding how rough they felt about the mid-air haptics stimulus, and if they experienced emotional responses (valence and arousal).Participants were asked to explore each audio-tactile stimulus one by one.A total of 9 stimuli were explored during the study.The whole process of the study procedure is show as in Figure 8.

Participants
In study 2, 18 participants volunteered (9 females, 9 males, mean age ± SD: 28 ± 7.8 years, range = 19-51 years), none of whom reported any impairments afecting their perception of mid-air haptics stimuli (e.g.neuropathy or vascular problems).All participants were recruited from the university's participant pool and were compensated for their time with a gift voucher of approximately $12.5.In the study, 4 participants had no prior experience with mid-air haptics devices, 9 participants had limited experiences, and 5 participants had extensive experience with mid-air haptics devices.Among all participants, only 4 of the participants indicated that they took part in the prior user study of this present paper.

Questionnaire analysis and results
Using questionnaire ratings, we validated how audio-tactile stimuli afected tactile and afective responses to mid-air haptics fabric textures based on roughness, valence, and arousal.Then we compare this results with our interview analysis in the next section.
In order to assess whether audio input can afect touch perception of the mid-air haptics texture stimulus, we used the ratings for A1-No audio as a baseline.A comparison of the ratings of fabric rubbing sounds (including stimuli A2, A3, A4, A5, and A6) with the ratings of the baseline condition A1 was conducted.Specifcally, our baseline ratings for stimulus A1-No audio were compared with A2-Original, A3-Frequency of -12dB, and A4-Frequency of +12dB frst.Following by comparing A1-No audio with A5-Volume of 200% and A6-Volume of 50%.Finally, we compared the baseline condition against the three pure tone sound conditions of 100Hz, 450Hz, and 900Hz.When necessary, sphericity correlations are applied, and Bonferroni/Holm corrections are applied for multiple comparisons.Below, we present our analysis and results in detail.

Roughness. In terms of the baseline condition (A1-No audio)
and the A2-Original, A3-Frequency of -12dB, and A4-Frequency of +12dB.There was no signifcant efect of audio on roughness ratings ( (2.335, 42.0) = 1.377, = .264,2= .071).The next analysis compared the baseline condition (A1-No audio) to the high (A5-Volume of 200%) and low volume (A6-Volume of 50%) conditions, again reporting a non-signifcant efect with a negligible efect size ( (2, 36) = 0.099, = .906,2= .005).Finally, there was no signifcant diference between the baseline (A1-No audio) and pure sound frequency conditions of A7-100Hz sound, A8-450Hz sound and A9-900Hz sound ( (3, 54) = 1.639, = .191),although there was a medium efect size between groups (2= .083).The results of the above analysis are show in Figure 9 (a).Overall, there was a medium efect size, despite no signifcant efect in roughness ratings.The reason for this results may be that a) we used a mixed method approach that include interview data, which meant that there may not have enough subjects to show a signifcant efect size; and b) there is a very small variation in the roughness perception of the fabric's mid-air haptics texture, because we used only one mid-air haptics texture from a very smooth and soft fabric (based on our results in Study 1 see 3.5).

4.4.4
Qestionnaire results summary.Although there are no significant efects in the ratings of roughness and valence, the ratings for the emotional ratings of arousal (calm or exciting) displayed a significant efects.Participants tends to have higher arousal ratings when they experience the audio-tactile experiences of the mid-air haptics fabric texture stimulus A8-450Hz sound and A9-900Hz sound.
In spite of the limited signifcance found in the questionnaire analysis, our qualitative analysis of the interview presents a diferent point of view.

Interview analysis and results
Here, we present the results from the qualitative analysis, which we then compare with the quantitative analysis to obtain a more comprehensive understanding of the user's audio-tactile perceptions.Both interview transcripts and notes taken during all interviews by the main researcher were used in the analysis.The collated data were frst coded and a series of themes were identifed following a thematic analysis approach [6], and then we used a discursive psychology analysis to explore and unpick some of the underlying language themes within the fndings [14].
The interview data revealed valuable insights into participants' perception of the mid-air haptics stimuli through the addition of sound.All participants were amazed by the variation in tactile perception with diferent audio input, although only one mid-air haptics texture was used.This stands in contrast to our questionnaire results, highlighting the importance of a mixed-method approach to understand the fne-granularity of texture and roughness perception among the nine audio-tactile stimuli.Below, we present each of the themes as provocative questions in detail.

4.5.1
It felt rougher, or smoother?All participants expressed to experience an increase in perceived texture roughness as the pure tone sounds of 450Hz and 900Hz were played while they explored the mid-air haptics texture stimulus.As an example, one participant described stimulus A8 (the mid-air haptics stimulus with 450Hz pure tone sound) as rough, as there was fast movement on the hand [P013].In contrast, a mid-air haptics texture with the rubbing sound of fabrics was perceived by participants as a smoother texture.Especially with the reduced audio frequency -A3 (-12dB).Participants described this experience "like moving hand in a water" [P003] and "it is the smoothest for so far" [P006].Additionally, when sound was manipulated at lower frequencies (A3)/volume is decreased (A6), participants tended to have a smoother texture perception.For example, one participant expressed after the exploration of the stimulus A3 (-12dB): "It feels very very smooth...it remind me of those printing papers...very smooth textures" [P001].For the stimulus A6 (volume 50%) participants clearly expressed that the experiences felt "smooth but also rough" [P006], kind of "smoother but stronger" [P018].4.5.2Associations to past experiences, or objects?It was common across participants to describe their audio-tactile perception in relation to an object or past experiences.There is, however, a tendency for the associations to be afected by diferent modalities depending on the audio inputs.One participant explained this association explicitly during the interview after they explored the stimulus A7 (100Hz), he/she expressed the stimulus reminded of a flm experience and added: "[the] association transports [you] through your imagination with the external stimulus" then they added "... it is the sound afected [me] a lot" [P018, A7].
During the exploration of 100Hz, 450Hz and 900Hz pure tone sounds participants often associated their audio experiences with past experiences or objects.We observed how participant P007 describes their experience with A8 (450Hz), saying it feels like an "electric stimulus" because it "sounds like electric".In contrast, participants are more likely to associate the rubbing sound of the fabric with their past experiences or objects, based more on their tactile experiences.A typical example would be when participant P002 described their perception of stimulus A4 (+12dB) as: "it feels like touching a waterfall" [P002].4.5.3Feels like fabrics, or not?It was more likely for participants to associate fabric experiences with the touch-related sounds (audios of rubbing fabric) than with pure tone sounds at 450Hz (stimulus A8) and 900Hz (stimulus A9).Audio-tactile experiences paired with a decrease in fabric rubbing sound frequency (A3) and volume (A6) were often associated with light and soft materials.Participant P016, for example, described the experience of stimulus A3 (-12dB) as being similar to the feeling of touching a "light and thin silk material".Participant P001 expressed that the mid-air haptics texture stimulus with the decreased volume (A6) reminded them of soft fabric, such as a "fur-coat".Further explanation was given by relating it to the experience of "stroking a cat fur" [P001].Soft fabrics, such as suede jacket material, also correlated with 100Hz sounds (stimulus A7), according to participants [P001].However, participants often found that "the audio is more dominating" [P010].Participants tended to associate music instruments rather than fabrics with audio-haptic experiences of pure sound at 450Hz (stimulus A8) and 900Hz (stimulus A9).4.5.4Feels nice, warm or not?The audio-haptic experiences of participants displayed a clear trend during the entire study, that fabric rubbing sounds (stimulus A2, A3, A4, A5 and A6) were more calming, while pure sounds at 900Hz (A9) were more disturbing.There were only two participants who expressed emotional responses after interacting with the mid-air haptics texture stimulus without audio input (stimulus A1).
Participants often described their experiences with the mid-air haptics texture stimulus combined with the reduced frequency of fabric rubbing sound (A3) as "relaxing [P017]" or "makes me feel calmer" [P014].Even with the unchanged audio of the fabric rubbing sound (stimulus A2), participants expressed calmness towards the entire experience, with participant expressing that the experience "remind me of lying on the grass and touching it by my side at night" [P002].Fewer people expressed emotional responses to the midair haptics texture without audio input (stimulus A1) than to the one with audio input.A typical emotional response to the midair haptics texture stimulus without audio was "no feelings" or "I focused more on the hand without sound" [P016].
In contrast, participants tend to express their emotional responses to the mid-air haptics stimulus with pure tone sound frequencies of 450Hz (stimulus A8) and 900Hz (stimulus A9) as "alerting" and "annoying".Participants often found the stimulus "not pleasant" compared to the rubbing sound of fabric.However, with the pure tone sound of 100Hz frequency (A7), participants found the experience more likely to be "pleasant" and "calming".
Interesting to note, in spite of using the same mid-air haptics stimulus throughout the study, participants expressed a feeling of temperature.Participants typically expressed warmth after they began to associate audio-tactile perception with fabrics or fabric products.For example, one participant expressed that stimulus A5 (increased volume) felt warm, soft, and smooth which they associated with a woollen scarf.Contrary to this, participants reported feeling warmth on their hands due to the mid-air haptics texture combined with pure audio at 100Hz (A7).The temperature sensation on the hand was described as "I don't know why but this one feels quite warm on the hand" [P018].

Study 2 summary and discussion
Although limited efects were found in the questionnaire data, our interview data showed a clear trend in the roughness perceptions among the nine stimuli.Participants tend to feel a rougher texture with the sine wave of pure tone sounds of 450Hz and 900Hz.The touch-related sounds with reduced frequency (A3-Frequency of -12dB) and decreased volume (A6-Volume of 50%) tends to produce a smoother texture perception compare to other stimuli.
In comparison to the emotional responses of valence (unpleasant or pleasant) and arousal (calming or exciting) collected through questionnaires, the interview results are similar.In other words only a few expressions were related to emotional responses.In response to the question: "how do you feel about the stimulus" or "no special feeling", a common response was "not much feeling" or "no special feeling".Worth noting that a high percentage of participants displayed fewer emotions when audio input was absent (stimulus A1-No audio).Furthermore, participants were feeling more calm when presented with rubbing sounds of fabrics (stimulus A2, A3, A4, A5 and A6).In stimuli containing pure tone sounds of higher frequency (stimuli A8-450Hz sound and A9-900Hz-sound), participants found the experiences more "alerting" and "annoying".
It appears, however, participants tends to express a wider afective experiences while presented with audio-tactile stimuli than solely mid-air haptics.During the interview, participants' association tends to biased by their tactile experiences when the touchrelated sounds are played.In contrast, the association to the past experiences and objects tends to be biased by the sound when the pure tone sounds was embedded.Interview analysis shows there is a higher tendency of expressing more association to their past experiences more when the audio feedback of rubbing sound of fabric played back to them.It seems that participants also more likely relate the audio-tactile experiences with fabrics when the fabric rubbing sound was played back to them than when the pure tone sound was presented.However, when participants experienced the pure tone sounds of 100Hz, 450Hz and 900Hz, they seemed more attracted to the auditory cure than the tactile stimulus.One participant explained it as follows: "My attention was attracted by the sound...it's the sound that afects a lot, I think" [P018].
While participants expressed a feeling of warmth when exploring the mid-air haptic texture with pure audio at 100Hz (A7), they did not perceive any additional thermal sensations.This may be due to three factors: a) low frequency sounds may evoke feelings of warmth when combined with mid-air haptics; b) these thermal experiences may arise from mid-air haptic operation; and c) individual thermal experiences may difer.Since participants only expressed warm experiences with 100Hz pure sounds audio-tactile stimuli (A7), it is likely that low frequency pure sounds, combined with mid-air haptics, could elicit warm feelings.Further studies are required to determine how pure tone sounds can be used to evoke a wide range of thermal sensations for mid-air haptic textures.

DISCUSSION
In this work -FabSound, we present two studies on perceptions of texture, roughness, and afective responses to mid-air haptics fabric textures.It appears that audio-tactile experiences of digital fabrics have a great potential to alter users' perceptions of roughness compared with the results we discussed in Studies 1 and 2 (see Section 3.5 and Section 4.6).As part of our discussion in this section, we explored the possibility of extending haptic interaction to multisensory experiences for digital fabrics, as well as proposing future applications and scenarios.

Towards multisensory experiences design
In this study, we frst examined the technology -mid-air haptics texture, and then combined tactile and auditory cues, presented a work that explored users' audio-tactile experiences of the digital fabrics.According to our results, mid-air haptics is capable of mapping fne detail surface textures, such as fabrics.The users, however, tend to perceive the mid-air haptics fabric texture as a smoother textures.Participants in our study are able to fnd the diferences in roughness, but they often have difculty relating them to the actual samples of fabric.We found that mid-air haptics can recreate smooth fabric textures that were comparable to soft physical fabric textures.
Although evidence show that physical fabric's sound plays a signifcant role in determine the sensory comfort of fabrics in the textile research [9,57], little is known about digital fabric texture (such as mid-air haptics fabric texture).With our investigation in the digital fabric texture of mid-air haptics, we demonstrate that when auditory input is incorporated, the users' perception of the mid-air haptics textures may difer.Pure tone sounds can result in a rougher mid-air haptics texture when compared to touch-related sounds.In addition, our results in study 2 support previous works by Guest et al. [23] and Jousmäki and Hari [27] that, with a reduced frequency or volume, users' roughness perception can be altered to perceive a smoother texture.
Prior research suggests that users often consider the sound of fabrics as an attribute to determine the tactile comfort of fabric products [31].The perceptions of auditory and tactile information can signifcantly infuence the consumers' purchasing preferences [31].This infuences to the users' purchasing decision can now be extended to the digital world.As highlighted by Velasco and Obrist [51], the convergence of technology and senses is leading to new experiences that will redefne how we design interfaces and interactions in the future.Our fndings on the audio-tactile perception of mid-air haptics fabric texture roughness have a potential to extend the existing haptic technology design space, fostering multisensory experience design around fabrics.

Towards afective fabric experiences and applications
According to Xue et al. [57], fabric tactile experiences can trigger a range of afective responses from users.However, when the mid-air haptics texture is used alone, the results vary (see section 3.5).As an interesting comparison, our study 2 shows audio-tactile perceptions of digital textures can produce arousing experiences that involve broader afective responses.Specifcally, when we compared pure tone sound audio-haptic stimuli with fabric rubbing sound stimuli, we found that touch-related sounds were more capable of conveying tactile experiences related to materials, whereas pure tone sounds could convey past experiences of sound based experiences.Furthermore, we found that audio and mid-air haptics textures may be able to convey temperature feeling of warmth using pure tone sounds of 100Hz, which is a promising direction for future research.
Earlier research suggests we are entering a phase of digital communication where we are moving from "way of seeing" to "way of feeling" [43,51].As we demonstrate with FabSound, the haptic experiences of digital fabrics can be extended to a multisensory path with an increase in afective responses, enriching the feeling of the digital experiences.
With FabSound, we can now envision multisensory applications that incorporating audio-tactile and afective experiences in digital space.In the following part, we explore how these efects can be integrated into application scenarios, leading to future research.

E-commerce.
E-commerce is a highly efective way to provide products that meet consumers' personalized needs, and consumers are increasingly motivated to engage in product customization [31].The fndings of this study can be applied to this E-commerce scenario.In particular, the audio-tactile experience of mid-air haptics textures could assist users in personalizing their e-garments and enhancing their afective responses, resulting in a more engaging experience [31].Using the interface, consumers can choose product attributes and select options to create a new product online.E-commerce interfaces can be designed to tweak consumers' touch perception by adjusting the volume of audio-tactile mid-air haptics texture perceptions.For instance, with the sound of rubbing a velvet fabric, users can feel velvet dresses before they buy them.Users may then switch to a 450Hz pure sound to convey the roughness of a rain jacket that feels synthetic.With the use of multisensory stimuli, e-retailers can present their products in a way that diferentiates them from the competition [31].

Digital signage.
With the evolution of digital signage, it has become increasingly multisensory, touchless, and interactive [10].These interfaces are used primarily for public engagement and advertising, so they serve as portals through which people can interact with 3D digital content without wearing a headset [10].In such a context, we can envision creating a public digital signage that embeds audio-tactile experiences to enhance users' multisensory experiences and increase their motivation to engage with the content.With this new digital signage, consumers' touch perception can be altered via frequency adjustments to their audio-tactile experiences.

Design tools.
A wide variety of tools and game engines are now available to designers and developers to make their games more interactive, intelligent, and multi-sensory [54].Unity and Unreal engines, for instance, have plugins and assets that allow users to set up haptics output devices [48].We propose here that in light of the results presented in this paper, user experience designers could beneft from a design toolkit that can provide recommendations on possible audio-haptics stimuli that can improve users' texture judgments.Designers and practitioners can now make informed decisions about how to fne-tune the user's perceived mid-air haptics texture roughness by changing the frequency or volume.
Our exploration of the mid-air haptics fabric texture and its interaction with audio-tactile perception has revealed a wealth of intriguing insights.We perceive texture roughness in a variety of ways, infuencing not only by our tactile experiences, but also by our auditory cues.Integrating mid-air haptics technology and auditory cues has provided a promising avenue for understanding how these elements infuence our perceptions of texture roughness.

Limitations and Future Work
This present work provided valuable insights into audio-tactile experiences with mid-air haptic fabric textures.In light of the increasing interest in haptics and digital touch, FabSound provides initial insights into how to improve the perception of digital through a multisensory lens.Future work should take into account auditory cues as well as other sensory cues to improve digital fabric tactile experiences.Furthermore, it is interesting to fnd that participants reported the feeling of warmth on their hands when the mid-air haptic texture was presented with pure tone audio at 100Hz (as discussed in 4.6).Further research can now investigate into how different pure tone sounds presented alongside mid-air haptic textures could elicit diferent thermal experiences.Moreover, as discussed in section 3.5, our study 1 showed that participants were more likely to fnd mid-air haptic texture stimuli to resemble smooth surface than rough fabric samples.This may be due to the limitations of current haptic mapping algorithms in mapping fne textured surfaces (such as fabrics).The development of a texture mapping algorithm that incorporates fabrics' physical properties may be necessary to overcome limitations in current haptic mapping algorithms.

CONCLUSION
Our work, FabSound, investigated whether mid-air haptics devices could be used to map fne detail textures fabrics, as well as the degree of similarity between the physical fabric samples and the mid-air haptics texture samples.Following by a study investigated how auditory perception of pure tone and touch-related sound could alter perceptions of mid-air haptics texture stimuli in terms of roughness.Our results demonstrated that physical fabric samples can be mapped using mid-air haptics devices.Although users found it hard to identify the matching physical and digital textures, users were able to fnd diferences in roughness between the mapped 9 mid-air haptics fabric textures.Moreover, while pure tone audio of 450Hz and 900Hz increases the perception of roughness, the sound of rubbing fabric have the potential of improving the association to fabrics.Additionally, our emotional ratings of arousal suggest that texture roughness audio feedback is more calming than pure tone sound feedback.Pure feedback rating, on the other hand, is signifcantly more arousing than audio feedback based on texture.In combining auditory and haptic experiences for mid-air haptics feedback, we demonstrate that multisensory experiences can enhance mid-air haptics experiences for fabric textures, enabling the development of richer mid-air haptics interfaces.Furthermore, we envisioned future interfaces, including E-commerce, digital signage and design interfaces, that combine two modalities to ofer a more immersive experience, as well as providing a more engaging user experience.

Figure 2 :
Figure 2: (a) Images of the selected nine cotton fabric texture images to the displacement map images and to the normal map images; (b) summary of mid-air haptics fabric texture mapping process: from textile material (fabric) to textile image and to mid-air haptics textures.

Figure 3 :
Figure 3: Figures shows the summary of the study setup and the procedure: (a).fabric sample of 20 × 20 cm was mounted on A3 white card; (b). the full setup of the Study 1: where participants can explore fabric samples to their left and mid-air haptics texture to their right; (d).summary of study process: each mid-air haptics stimuli was compared to nine fabric samples, and the questionnaire was flled out.This process was repeated nine times for nine mid-air haptics stimuli in total.

Figure 4 :
Figure 4: Summary of similarity rating analysis of nine mid-air haptics texture (H1-H9) × nine fabric samples (F1-F9): each of the fgures from (a) to (i) present the result of each similarity ratings of stimulus (H) and the fabric samples (F1-F9).For example, plot (a) (top left) illustrates similarity ratings between the digital haptic of Heavy Drill (H1) and the nine physical fabrics (F1 -F9), with the analysis indicating that F3 (Buckram) was rated as less similar to the digital haptic of Heavy Drill (H1) relative to Muslin (F8) and Twill (F9).Left side plots represent the mean and standard deviation with right side plots representing the total data distribution.Note: * = p < .05,** = p < .01.

Figure 5 :
Figure 5: Results of roughness ratings of Study 1: a statistically signifcantly higher roughness ratings for H2-Digital Flannel than H6-Digital Voile and H2-Digital Flannel than H7-Digital Calico.Left side plots represent the mean and standard deviation with right side plots representing the total data distribution.Note: * = p < .05,** = p < .01.

4. 1 . 1 1 .
Audio-tactile texture stimuli.Study 2 employs a mid-air haptics texture stimulus mapped from F5-Velvet, with its recorded touch-related sound (the rubbing sounds of the velvet) and the generated pure tone sounds.Overview of this process is show in Figure6(c).Using the following methods, we obtained 8 diferent auditory recordings: Original recording of the fabric rubbing sound: To record the touch-related sound, we rubbed the fabric F5-Velvet and record the sound using a Rode NT2 A Studio Condenser Microphone connected to a Focusrite Scarlett 2i2 Studio Audio Interface (frequency response 20-20kHz±0.06dBand gain range 69dB).The Rode NT2 microphone was placed close to the fabric.The fabric was then rubbed in one direction with constant force and speed in a circular movement for 20 seconds.This movement allows us to record the fabric rubbing sound with minimum hand exploration frequency.The recording setup is shown in Figure 6 (a).Throughout the recording, the audio gain of the Focusrite audio interface was set to its maximum.The recorded audio was then imported into an audio software Audacity (Version 3.4.2) and generate a Mel-Frequency(MEL) spectrogram, as shown in Figure 7 (a).2. Frequency changed sound: In Adobe Premiere Pro 2023 (23.6.0), the recorded fabric rubbing sound's frequencies was decreased by -12dB and increased by +12dB.Following that, we exported three audio fles with the original frequency, increased frequency, and decreased frequency of the rubbing sounds.Each of the audio fle was then imported into an audio software Audacity (Version 3.4.2) and generate a Mel-Frequency(MEL) spectrogram, as shown in Figure 7 (b) and (c).These three audio fles will then be merged with mid-air haptics textures.3. Volume changed sound: A video editing software (IMovie, version 10.3.6) was used to change the original fabric rubbing sound recording volume from 100% to 50%, and from 100% to 200%.The volume-changing sounds were then exported for further processing.Each of the audio fle was then imported

Figure 6 :
Figure 6: (a).Image of the setup for recording the rubbing sound of fabric F5-Velvet; (b).Image of the full setup of Study 2; (c).Process of creating an audio-tactile mid-air haptics experiences for Study 2.

Table 2 :
Details of Audio-haptics Stimuli in Study 2: each of the stimulus was assigned to a number from A1-A9 Stimulus (A) Detail of audio-tactile stimuli A1 -No audio Digital Velvet only A2 -Original Digital Velvet + Recorded fabric rubbing sound A3 -Frequency of -12dB Digital Velvet + Recorded fabric rubbing sound frequencies were decreased by -12dB A4 -Frequency of +12dB Digital Velvet + Recorded fabric rubbing sound frequencies were increased by +12dB A5 -Volume of 200% Digital Velvet + Recorded fabric rubbing sound volume were increased to 200% A6 -Volume of 50% Digital Velvet + Recorded fabric rubbing sound volume were decreased to 50% A7 -100Hz Sound Digital Velvet + sine wave pure tone sound of 100Hz A8 -450Hz Sound Digital Velvet + sine wave pure tone sound of 450Hz A9 -900Hz Sound Digital Velvet + sine wave pure tone sound of 900Hz

Figure 7 :
Figure 7: Mel-frequency (MEL) Spectrograms of (a) original recorded fabric rubbing sound; (b) recorded fabric rubbing sound with frequencies decreased by 12dB; (c) recorded fabric rubbing sound with frequencies increased by by 12dB; (d) Recorded fabric rubbing sound with volume increased to 200%; (e) Recorded fabric rubbing sound with volume decreased to 50%.