Authors: Christopher S. W. Anderson, Miles Thorogood
Categories: Original Article, Immersive audio, Virtual reality, Sound dispersion, Listener preference
Source: Virtual Reality
Authors: Christopher S. W. Anderson, Miles Thorogood
Spatialized sound is crucial for immersive and extended reality experiences. We conducted a user study comparing the built-in audio of a commercially available VR headset with a physical loudspeaker dispersion systems. Our study demonstrates participants have a greater sense of immersion, realism, and localization phenomena when experiencing the physical loudspeaker dispersion over the built in headset. A subsequent preference test and qualitative feedback revealed a strong trend favoring the both high-end and entry-level loudspeaker systems. This work provides valuable insights for designing more immersive extended reality experiences.
Immersive experiences in extended reality (XR) rely on the fusion of multisensory stimuli to create perceptually compelling spaces. In particular, sound plays a critical role in many Virtual Reality (VR) environments by localizing objects, events, and influencing the sense of space in a scene. Implementing affective sound localization systems for head mounted displays (HMD) that deliver strong spatial qualities is essential for users in perceptually convincing and satisfying experiences (Mavridou et al. 2025). In this paper, we outline our investigation to determine whether the use of a multichannel loudspeaker layout for VR creates a different experience compared to the 3D audio reinforcement system built into common headsets. The purpose of our study is to determine whether there is a perceived preference for VR HMD headsets or sound dispersion audio systems such as a loudspeaker array while in a virtual reality experience. To examine user preference, we conducted a listening study to determine the perceived preference of either a spatial VR headset’s built-in audio system or an audio dispersion system using loudspeakers in the physical environment. The study has three phases with different conditions. In phase 1, we investigated the qualitative results of the participant feedback after repeated listening to the audio from both systems. In phase 2, we conducted a preference test and analyzed the quantitative and qualitative results reported by the participants. Finally, in Phase 3. we substitute the high-end loud speaker array for an entry-level out-of-the-box spatial audio solution to measure preference over a range of systems.
Spatial audio is a sound design technique that enhances realism and immersion in virtual reality (VR) by replicating how humans perceive sound in a 3D environment (Gilberto et al. 2025). It allows sounds to appear as though they are coming from various directions and distances, enabling users to pinpoint their sources accurately within the virtual world. One widely adopted technology behind this is the Head-Related Transfer Function (HRTF), which simulates how sound interacts with the human head and ears to produce realistic 3D audio (Cheng and Wakefield 1999). HRTF models how sound from a specific location reaches the ears, accounting for factors such as the shape and size of the head, ears, and ear canal. HRTF typically requires the use of headphones to recreate these psychoacoustic properties. However, HMD VR headsets, such as Meta Quest 3 incorporate a universal HTRF method using a set of built-in directional speakers (Meta 2014).
The entertainment industries have adopted three-dimensional spatial audio standards and technologies for immersive sound in films, video games, and music. One widely accepted standard in this field is Dolby Atmos (DA), a spatial audio standard designed to enhance the listening experience by adding a height dimension to traditional surround sound settings (Pfanzagl-Cardone 2023). Unlike conventional audio systems that assign sounds to fixed channels, Dolby Atmos utilizes object-based audio, allowing sounds to be precisely positioned and moved within a three-dimensional space. Film studios use the technology to create lifelike cinematic soundscapes, while game developers utilize its spatial accuracy to heighten player engagement, and musicians are exploring its potential to communicate multidimensional music experiences. DA has the ability to adapt to various platforms such as headphones, professional, and consumer-grade multichannel systems (Sergi 2013). Our multi phase study described in this article seeks to identify user preference between spatial audio delivered by headsets and multichannel loudspeakers to provide directions for future immersive VR and XR audio experiences.
In this section we survey the state of the art on immersive sound research for VR. As spatial audio comparison methods are varied, this section is divided into 5 categories of configurations that are used to study virtual audio spatialization. The categories Loudspeakers, Headphones and Loudspeakers (No HMD), HMD with 360 Degree Video, HMD with 3D Visuals, HMD and Loudspeakers with 3D Visuals.
To situate the use of loudspeakers in our research, we have examined other spatial audio studies that explore audio for VR or virtual audio spatialization preferences using only loudspeakers. Identifying and determining important perceptual attributes in reproduction systems is an important step to understand potential listener preferences, as indicated in the studies of Francombe et al. (2016, 2017), which elaborate on perceptual attributes that are connected with listener preference. Of these, they discuss how Choisel and Wickelmaier (2007) used a set of eight width, elevation, spaciousness, envelopment, distance, brightness, clarity, and naturalness of which they collected ratings and listener preferences for eight reproduction methods from mono to surround sound in a paired comparison test.
Studies of virtual spatial audio often combine the use of headphone and loudspeaker technologies to compare spatialization methods such as Dolby Atmos and Ambisonics. These comparison studies expand on the details of audio processing methods and examine the accuracy of sound localization in relation to listener perceptions and experiences. For example, Malecki et al. (2023) conducted a comparison study of spatial audio mixing of Ambisonics and Dolby Atmos. The study involved a series of listening tests that revealed insights into the efficacy of these technologies. The researchers mention that despite the Ambisonic mix scoring higher in some quantitative areas, subjective tests indicated a general preference for the mix produced by Dolby Atmos technology in all listening systems regardless of loudspeaker setup and was likewise observed in binaural listening. Researchers such as Thresh and Kearney (2017) have conducted similar studies that compare the localization performance of first, third, and fifth-order Ambisonics to listen to loudspeakers and headphones in virtual settings by simulating the acoustics of a room using binaural impulse responses. Riedel and Frank (2024) also compared non-individual binaural audio with loudspeaker reproduction by investigating perception of distance and elevation, as well as listener envelope and engulfment. They used open headphones, which enabled the use of real and virtual sound sources. Mróz and Kostek (2023) studied perceptual responses in audio-visual interactions of binaural spatial audio where participants were tasked to watch a screen and indicate the direction of the incoming sound while listening to the audio via loudspeakers or headphones with an HRTF plugin. Furthermore, Huisman et al. (2019) used a headset with a loudspeaker array to study the location, timing, and distance of sound between audio and visual stimuli to establish a baseline for hearing studies.
Other audio research studies have used HMDs to assess spatial audio within immersive audio visual contexts. Much of this research has focused on the use of audio for 360 degree video immersion. An example of this would be the research by Hong et al. (2019) on the quality of spatial audio reproduction methods in VR to determine their ecological validity for soundscape research. The researchers compared three different spatial audio reproduction methods in VR: binaural, loudspeaker array, and ambisonics. Their paper investigated the impact of different factors on the perceived quality of spatial audio in VR, such as the type of sound source, the level of immersion, and the degree of realism. The findings of their paper suggested that VR has the potential to be a valuable tool for soundscape research, but that the choice of spatial audio reproduction method can have a significant impact on the perceived quality of the soundscape. Similarly, Davies et al. (2017) investigated audio models for virtual reality (VR) video and the sense of immersion each model delivered. They aimed to determine the perceived effects of various sonic presentation modes in a VR space where participants experienced a VR video with four different audio Mono, Stereo, 5.1 Surround Sound, and a Virtual Spatial Position configuration.
VR spatial audio researchers have also focused on studying interactive 3D experiences. Research in this area is of importance towards conducting VR headset evaluations when multiple sensory inputs are involved. For example, Robotham et al. (2022) investigated the effect of six-degree-of-freedom sensory VR (6-DoF) on the quality ratings of real-time audio rendering. They conducted a comparison of a series of evaluations to identify audio qualities within a multi-modal VR context. They imply that complex scenes and interactive aspects of 6-DoF VR can impede quality judgments. They also discuss how the subjective quality of evaluation for media technology is dependent on the context of use and the demographics of the users. They point out that traditional VR audio quality evaluations have mostly considered a context where the audio is judged as an independent sensory input as opposed to multiple sensory inputs. In a large scoping review of other HMD focused audio research, Bosman et al. (2024) mention that the effects of audio in various contexts such as this are still poorly understood and that research in this area still contains inconsistencies in measuring outcomes and selecting stimuli. Their scoping review explored 121 studies on the effects of audio on player experiences in HMD-based VR and found that the most common areas of the studies were within entertainment, soundscapes, therapy/rehabilitation, and medium-specific areas. They also found that presence, realism, immersion, and pleasantness were found to be the most widely studied experiential aspects.
Some researchers have combined VR HMD audio with 3D imagery and loudspeaker technology such as Wu et al. (2024) who conducted a comparative study on the impact of auditory cues on balance performance. Their study examined the influence of synchronized foreground sounds and background distracting sounds on participants’ balance while navigating a VR environment. By investigating the effectiveness of these two sound delivery methods, the research was aimed at contributing to the development of human movement and balance rehabilitation methods. Some researchers have also compared user preferences of HMD headphone and loudspeaker audio. One such study was conducted by Marais and Foss (2023) who performed a comparative analysis of VR audio based on headphones and speakers for games. Their goal was to find the differences between immersive spatial audio implemented using loudspeakers and headphones in VR applications. They implemented a loudspeaker and a headphone spatialization system, but like other studies they mostly focused on gathering quantitative data on localization and positional accuracy. Their quantitative results revealed that the loudspeaker-based system achieved a lower average positional accuracy than the headphone-based system and found that the variance in positional accuracy of both systems was not statistically significant. However, the qualitative component of their study that included preferential responses indicated that despite a difference in perceived audio quality between both systems, 59 percent of the participants preferred the speaker environment. The loudspeaker preferences of the participants were said to be “more natural”, “less claustrophobic” and “more immersive”. They also suggested that these results introduce a significant alternative to headphone-based audio for VR HMDs.
The related research discussed in this survey section has largely focused on obtaining specific results from a variety of immersive sound localization studies, while others have explored user experience and preference based on specific spatial audio dispersion configurations outside of the scope of game settings (Mcarthur et al. 2023; Rumsey 2020). Schoeffler et al. (2017) have assessed basic audio quality (BAQ) and overall listening experience (OLE) of 3D immersive audio using two-channel stereo, 5.1 surround and 22.2 3D material, although they did not assess preferences between Dolby Atmos and the HMD headset in a game setting. An earlier study by Schoeffler et al. (2015) explored the OLE in a VR environment compared to the real-world environment of a cinema and music studio. However, commercially available VR Headsets and immersive audio technology standards have advanced since their study. Hedges et al. (2024) have since surveyed newer studies that have explored immersive audio in XR. They found that only (n=12) of the studies incorporated loudspeakers while headphone studies were covered in (n=67) studies. However, none of the studies in their paper covered the spatial audio from HMD headset-based speakers combined with loudspeaker audio dispersion systems. As such there is a knowledge gap in the comparison of user preferences for Dolby Atmos in this configuration. Hedges et al. (2024) also mention that there is a need to broaden the scope of the field to include more research in immersive audio in AR and MR environments. This aligns with the introduction of AR technologies such as XReal AR glasses which include a set of Bose speakers similar to the speaker setup in the Meta Quest 3 headset (XREAL 2026).
This research investigates user preference for immersive audio delivery in VR, focusing on both HMD headset-based and loudspeaker-based dispersion systems. Our methodology used a three-phase empirical first, to characterize the affective impact of distinct immersive sound systems (Phase 1), second, to quantify user preference between these systems (Phase 2), and finally (Phase 3) - to measure preference over a range of systems.
Phase Affective Immersion Profiling. This phase explores the subjective experience generated by each audio system within a controlled VR environment. We detail the specific virtual environment dataset and the configurations of the immersive sound systems under evaluation. Participants participated in a survey, providing rich qualitative data on their perceptual experiences. They were asked to describe their experience with the in-built VR spatial sound system as well as the loudspeaker system. They were asked what they liked or disliked about each setup.
Phase Comparative Preference Quantification. Based on the affective profiles derived from Phase 1, Phase 2 directly measures user preference between HMD and loudspeaker systems. The participant responses to an A / B preference test are analyzed using a two-tailed Binomial test, providing statistically robust information on the magnitudes of the preferences.
Phase Comparative Preference with entry Level Spatial Audio System This section modifies the conditions of the experiment by substituting the high-end loudspeaker array in Phases 1 and 2 for an entry-level spatial audio system of quality similar to the headset audio technology.
As our study involved human participants, we obtained approval from the University of British Columbia’s Office of Research Ethics (UBCORE). In accordance with their ethics review, we confirm that all experimental protocols were approved prior to the start of our investigation, and all methods were carried out in accordance with the UBCORE guidelines and regulations. Before data collection for each phase of our study, informed consent was obtained from all participants.
We used two immersive audio systems readily available to consumers. We first sought to understand the affective qualities of two different immersive audio systems. This section describes our methodology for the first qualitative study phase of our research.
For our research purposes, we utilized and modified an instance of Audiokinetic’s Wwise Audio Lab (WAL) level. WAL is an open source 3D game environment level developed for the Unreal Engine video game development platform to demo the spatial audio capabilities Audiokinetic’s widely-used game sound middleware Wwise (Audiokinetic 2025). The foreground sound assets in the WAL are placed as virtual objects within the 3D environment, which are attenuated in relation to the distance of the player camera. This enables the panning and attenuation based on the position of the head tracked six-degrees-of-freedom (6DOF) movement of the HMD. The background sounds are positioned as surrounding zones or reverb areas. Some of the objects in the environment use a level-of-detail (LOD) approach to change the characteristics of the audio and are configured to use foreground and background sounds to increase or decrease the amount of sound layers based on distance attenuation curves. LOD also applies to some effects such as filtering or reverb levels. The WAL virtual environment simulates a first-person representation of an island village with various listening locations during the day. For Phase 1, we used the following 5 listening locations that contain a variety of foreground and background Standing in an abandoned cathedral where the user is isolated and sounds are mostly ambient.Foreground owls, wind, bird wings flutteringBackground cathedral room tone ambient, distant bird wings fluttering, wind, filtered sounds from outsideStanding in an abandoned cathedral in front a large room with people inside talking. Their audio can be heard as reverb through a set of doors which act as a portal.Foreground owls, wind, bird wings flutteringBackground distant muffled talking in a separate room, cathedral room tone ambient, distant bird wings fluttering, wind, filtered sounds from outsideStanding at an outdoor park in front of a spinning water fountain.Foreground rotating water flow, water splashesBackground seagulls, wind, ambient distant ocean wavesStanding at an outdoor campsite in front of campfireForeground Multiple assets of fire crackling that fade in or out based on LOD.Background seagulls, wind, ambient distant ocean waves, distant water fountain.Standing at a seaside cliffForeground ocean waves crashingBackground seagulls, wind, distant ocean waves
The first condition was an off-the-shelf built-in spatial sound system. In this arrangement, the user wears a typical VR headset (Meta Quest 3) with small audio drivers located on the headset headband above each ear.
The second condition used a multichannel loudspeaker array setup in the listening space surrounding the user. The specific location of each loudspeaker was calculated based on the Dolby Atmos 7.1.4 specification (Novotny 2024).
For both of these conditions, we configure a customized version of WAL simultaneously on a Windows PC with an NVIDIA RTX 3090 GPU, which uses wireless tethering and Meta Quest Link to reduce audiovisual latency on a Meta Quest 3 HMD, and on an M2 Mac Studio Max for controlling the project setup, as well as the loudspeaker array over an audio video bridging (AVB) network. In WAL, we enabled remote switching between the HMD and loudspeaker array setups and ran the level simultaneously on the PC and Mac Studio by means of a multiplayer game session over a local area network (LAN) using a router. We used a custom program built in MaxMSP to control both multiplayer instances of the WAL level on each computer. This program ran on the Mac Studio and used Open Sound Control (OSC) messages to switch the audio configuration between HMD and loudspeakers and for the virtual teleportation of the user’s listening position within the WAL level. The WAL environment was configured for Dolby Atmos 7.1.4 audio output on both the PC and Mac. While the PC utilized Dolby Laboratories’ Access software on the audio output to process binaural HRTF audio for the HMD, the Mac Studio sent 7.1.4 channels of audio to an array of 11 Genelec 8330A loudspeakers and a 7360A subwoofer, via an RME Digiface AVB and two MOTU 24A AVB interfaces. This configuration is outlined in Fig. 1. The loudspeakers were arranged in a Dolby Atmos formation within a large performance space with a listening position at centre, as shown in Fig. 2. The playback level for the loudspeaker array and headphones was calibrated using an Earthworks M50 measurement microphone placed at the listening position. Firstly, it was determined that 65dBA was an acceptable volume for the loudspeaker array. Secondly, the headset relative level was determined subjectively by an audio professional’s ears to match the perceived amplitude of the loudspeakers and this perceived level was captured.Fig. 1This image is a diagram showing the configuration and signal flow of phase 1
Each participant was briefed and positioned standing in the center of the listening space, facing the center loudspeaker. All 5 virtual WAL locations were presented in a randomized viewing order 1 min and 10 s apart. The first 30 s “A" period of the listening location presented the HMD binaural audio configuration. The subsequent 30 s “B" period presented the loudspeaker array followed by a short 10 s of silence, where participants were asked for their preference A, B or neither. After visiting all 5 locations, the participant responded to the questionnaire.Fig. 2This image represents the 7.1.4 loudspeaker layout in the large performance space with the listener centered in the soundfield, left & right loudspeakers, center loudspeaker, left & right surround loudspeakers, left & right rear surround loudspeakers, left & right top front overhead loudspeakers, and left & right top rear overhead loudspeakers. The subwoofer was positioned on the floor
Thirty participants (N = 30) were recruited from the undergraduate and graduate student population on the Okanagan Campus of the University of British Columbia, Canada. Recruitment was carried out through flyers posted on campus and directly through word of mouth. Participants were required to have an accurate frequency listening range at an average 65dBA amplitude level to ensure that they could effectively perceive auditory signals within the study. Other demographic data, such as age, gender, and field of study, were not collected to preserve the anonymity of the participants.
Participants self-reported their familiarity with virtual reality (VR) using a five-point scale. The distribution of VR experience was as Never: 13% (n = 4)Rarely: 37% (n = 11)Monthly: 10% (n = 3)Weekly: 27% (n = 8)Daily: 13% (n = 4)This distribution indicates a range of VR experience among participants, with the majority reporting infrequent or no previous experience with VR technology. This range of experience is beneficial for assessing the usability and accessibility of the system for a diverse user base.
Participants provided feedback on their experiences with both the built-in VR spatial sound system (headset) and the physical space dispersion system (loudspeakers). A thematic analysis of these responses revealed several key themes within the qualitative data (participant feedback): immersion, realism, sound quality, and spatial accuracy.
Headset Participants had mixed experiences. Some found the close proximity of the audio to the ears immersive (“The audio being in close proximity to the headset itself is always nice (maybe because used to VR headsets)"), and the dynamic spatial sound added to the immersion (“I liked how responsive the spatial sound was as I look around the environment... the dynamic spatial sound made the overall experience more immersive."). However, others felt the sound was “closed in" or lacked a sense of wide space.
Loudspeakers The physical loudspeaker system was generally perceived as more immersive, creating a greater sense of presence and scale. “I like the wide space, the birds flying over the top." “Immersive. I wanted to walk and explore more. I felt relax”. “The last scene with lake and sea was amazing with this system”. Some participants mentioned that the audio was “part of the environment" with the headset but “detached from the scene” with the loudspeakers, suggesting a more seamless integration with the visuals for the headset in certain cases.
Headset Some participants found the sound less realistic, lacking depth and detail. “I like the stereo sound quality of the headset but it lack details. Not clean enough”. “It feels fake, while physical one give a natural feeling of the environment and better sense of depth."
Loudspeakers The loudspeaker system was frequently praised for its realism and naturalness. “Just like being real - everything about it is great... highlighting directional audio is so amazing and most importantly immersive”. “The physical space dispersion system definitely sounded more real and less processed, closer to real experience of hearing..."
Headset Sound quality was a common concern, with participants describing the audio as“subpar”, “not clean enough”, “muffled”, “tight”, or having “frequency overlap”. “Audio quality can be a bit subpar at times; occasionally I could not hear anything whatsoever”. “1. about quality, feel like a little loud even it has the same volume, feel like the system directly hits my ear”.
Loudspeakers The loudspeaker system was often praised for its higher sound quality, clarity, and depth. “I really like sound quality and how it much clear with in-game/vr object”. “It was way higher quality, although the echo in the room was more prominent, which was sometimes confusing”.
Headset Participants’ experiences with spatial accuracy varied. Some appreciated the clear sound localization and responsiveness to head movements (“In-built system were very clear. I really like that the sound is centered with my ears. When I rotate, the sound moved with me”.), while others felt the sound was less precise or dynamic. “It sounded as if the sound were just happening with very little sense of space”.
Loudspeakers The Loudspeaker system was often described as providing a more accurate and natural sense of sound distance and placement. “Where things were making sounds in the virtual world was a lot more clear and sounded better and more clear in general”. “I liked the proximity effect of the objects”.
This section describes our methodology for the second phase of our study which used both quantitative and qualitative methods.
The WAL environment for Phase 2 used the same virtual first-person level as Phase 1 but instead only presented 3 possible listening locations. The following is a description of the 3 listening locations with foreground and background sounds for Phase Sitting in an abandoned cathedral where the user is isolated and sounds are mostly ambient.Foreground owls, wind, fluttering bird wingsBackground cathedral room tone ambient, distant bird wings fluttering, wind, filtered sounds from outsideSitting in an outdoor park near a spinning water fountain.Foreground rotating water flow, water splashesBackground seagulls, wind, ambient distant ocean wavesSitting next to an outdoor campsite in front of campfireForeground Multiple fire crackling assets that fade in or out based on LOD.Background seagulls, wind, ambient distant ocean waves, distant water fountain.
The first condition used the same VR headset (Meta Quest 3) as in phase 1 with small audio drivers located on the headset headband above each ear.
The second condition used a multichannel loudspeaker array setup in a closer configuration that surrounds the seated user in a smaller acoustically treated studio space. The specific location of each loudspeaker was also calculated based on the Dolby Atmos 7.1.4 specification.
For the conditions in Phase 2, we configured another customized version of WAL for a single Windows PC-based configuration for better accuracy and easier audio switching. The Meta Quest 3 HMD was physically attached to the PC using the USB-C link cable, and remote switching between the HMD and loudspeakers was controlled by the PC keyboard, which was also used to trigger user listening locations. The PC utilized Dolby Access software to process binaural audio for the HMD while the WAL was configured to use Audiokinetic’s ASIO plugin to make use of the RME Digiface AVB interface (see Fig. 3). The loudspeakers were arranged in a smaller treated studio space with a seated center listening position (demonstrated in Fig. 4). This smaller studio space was chosen for this phase to better control acoustics and tame wall reflections that can alter perceived locality. The seated position was also chosen to prevent participants from changing their proximity to the loudspeakers. The playback level for the loudspeaker array and headphones was calibrated using an Earthworks M50 measurement microphone placed at the listening position. Firstly, it was determined that 65dBA was an acceptable volume for the loudspeaker array. Secondly, the headset relative level was determined subjectively by an audio professional’s ears to match the perceived amplitude of the loudspeakers and this perceived level was captured.Fig. 3This image is a diagram showing the configuration and signal flow of phase 2
Each participant was briefed and seated facing the center loudspeaker at the central point of the listening space. Only 1 of the 3 possible virtual WAL locations was randomly selected and presented for 2 min. The presentation order of the HMD binaural and loudspeaker configurations was randomized with a 1 min listening time for each. When switching to the loudspeakers from the HMD audio, the positional data stream from the HMD—which controls the binaural HRTF sound positioning—is disabled to avoid panning the sounds on the loudspeakers. Participants were asked for their preference for the first listening experience or the second. Half of the participants were randomly chosen to observe an animated ball being shot out of a launcher at random intervals into the virtual space from the listening point. The sounds created from the ball bouncing off the WAL environment surfaces provided a variety of observable collisions to enhance the experience of foreground sounds and their subsequent reverberations in the virtual environment. After visiting one of the 3 locations, the participant then answered the questionnaire. Randomization of the 3 listening locations and 2 configurations was achieved by grouping 10 questionnaires into 3 groups for each location. Within these groups, 5 were assigned to hear the HMD audio first and the other 5 were assigned to hear the loudspeakers first. The 3 location groups were shuffled back into a group of 30 questionnaires. From this shuffled pile, 3 groups of 5 were chosen at random to observe the ball launcher and then reshuffled with the rest of the questionnaires. This random selection method ensured that wide coverage of scenarios could be observed while minimizing possible selection or listening biases.Fig. 4This image represents the 7.1.4 loudspeaker layout with the listener centered in the soundfield, left & right loudspeakers, center loudspeaker, left & right surround loudspeakers, left & right rear surround loudspeakers, left & right top front overhead loudspeakers, and left & right top rear overhead loudspeakers. Note that the subwoofer is not shown, where it was positioned during the study on the floor at the front left
A new set of participants (N = 30) from the undergraduate and graduate student population of the University of British Columbia was recruited with the requirement to self report normal hearing to ensure they could effectively perceive auditory signals within the study. Participants self-reported their familiarity with virtual reality (VR) using a five-point scale. In the case of phase 2 here, a larger number of participants had little to no experience with VR. The distribution of VR experience was as Never: 33% (n = 10)Rarely: 33% (n = 10)Monthly: 10% (n = 3)Weekly: 10% (n = 3)Daily: 13% (n = 4)
To assess the overall preference between the two audio configurations (HMD vs. Loudspeakers), the data from all 30 participants was pooled and subjected to a two-tailed Binomial Test against the null hypothesis of equal preference (50%/50%).Of the 30 total participants, 8 (26.7%) preferred the HMD and 22 (73.3%) preferred the Loudspeaker Array.The Binomial Test revealed a statistically significant preference for the Loudspeaker Array configuration over the HMD, Pvalue=0.0063.The 95% Clopper-Pearson Exact Confidence Interval for the true population proportion of preference for the Loudspeaker Array was calculated as [0.541,0.877].This result indicates that the observed preference split (22 Loudspeakers vs. 8 HMD) is highly unlikely to have occurred by chance if the two audio methods were equally preferred. Participants exhibited a strong and significant tendency to prefer the Multichannel Loudspeaker Array setup. This CI indicates that we are 95% confident that the true percentage of individuals who prefer the Loudspeaker Array falls between 54.1% and 87.7% (see Fig. 5). Because the entire interval is above the point of no preference (50%), this strongly supports the finding that the Loudspeaker Array is significantly preferred over HMD for spatial audio experiences in this study.Fig. 5Participant preference for the spatial audio configuration. Participants showed a strong preference for the Loudspeaker Array (73.3%) over the HMD Binaural Audio (26.7%). The error bar represents the 95% Clopper-Pearson Exact Confidence Interval for the proportion of participants preferring the Loudspeaker Array. The CI [0.541,0.877] excludes the point of no preference (50%), indicating a statistically significant difference (p=0.0063)
Participants provided open-ended feedback on their experiences with both the in-built VR spatial sound system (headset) and the physical loudspeaker system. Several recurring themes emerged, including immersion, sound quality, spatial accuracy, and comfort.
Headset Although some participants found the headset audio immersive, particularly due to its close proximity to the ears, others felt it created a “closed in" or “smaller space" feeling. “I loved the experience... but the feeling was more in an enclosed space-kind of feeling. So it kind of made it feel like a smaller space." Some also mentioned the immersive environment due to the relation between movement and “I liked the immersive environment because of its relation to movement and sound (spatial audio)”.
Loudspeakers The physical loudspeaker system was generally perceived as more immersive, creating a “bigger space" and a more “realistic" experience. “I also enjoyed this experience as it felt as I was in a bigger space with the physical loudspeaker system." “The physical loudspeaker felt more realistic because of the depth." “It felt like I was actually in the scene (sound-wise). The sounds were coming from around me."
Headset Several participants commented on the lower quality of the headset audio compared to the loudspeaker system, describing it as “less realistic”, “lower quality”, “a bit quiet," “tight," “muffled," or “flat." “The quality of the audio was clearly a lower quality... manageable to understand... but the quality I prefer was by the physical loudspeakers” “Dislike: sound feel a bit tight, for big virtual space like this.”
Loudspeakers The loudspeaker system was consistently highlighted for its superior sound quality, and participants noted its “robust," “deep”, “clear”, “clean” and “realistic” audio. “The sound was much more robust, deep, and realistic. Depth is a lot more realistic as well...” “The quality I prefer was the physical loudspeakers. It was vastly clearer and cleaner”.
Headset The participants had mixed experiences with spatial accuracy. Some felt it provided a better sense of space and directionality, while others found the sound localization less precise or dynamic. “I liked number 1 experience more, because I had a better sense of space and I could attempt to follow the sound direction better”. “The sound felt a little forced when spatialized. As if it was trying to catch up with me”.
Loudspeakers The loudspeaker system was often described as providing a more accurate and natural sense of sound distance and localization, with some noting the echo effect as enhancing realism. “The sound was clear and the sound distance felt more accurate to the image I was seeing...” “Speakers had more immersion, more echo effect was prominent liked that in the room”.
Whereas Phase 2 studied the user preference for HMD spatial audio and a professional quality loudspeaker array, Phase 3 aims to determine the user preference between the HMD audio system and an entry-level spatial audio system. In this phase, the same Virtual Environment Data Set stimulus as in Phase 2 is presented to participants.
Identical to the Phase 2 conditions, the first condition used a VR headset (Meta Quest 3) physically attached to the PC using the USB-C link cable, and remote switching between the HMD and loudspeakers was controlled by the PC keyboard. However, condition 2 modified the treatment by exchanging the audio reinforcement to the least expensive Dolby Atmos enabled 7.1 soundbar system available at the time of publication (< $300CAD).1 This system consists of a soundbar containing three 2 inch drivers, four surround speakers with 2 inch drivers, and a 4 inch subwoofer. Overall, the system has an frequency response of 65Hz - 18kHz. The location of each of the speakers follows the 7.1 arrangement specified by the manufacturer. Again, the playback level for the loudspeaker array and headphones was calibrated using an Earthworks M50 measurement microphone placed at the listening position. Firstly, it was determined that 65dBA was an acceptable volume for the loudspeaker array. Secondly, the headset relative level was determined subjectively by an audio professional’s ears to match the perceived amplitude of the loudspeakers and this perceived level was captured.
Another set of participants (N = 30) from the undergraduate and graduate student population of the University of British Columbia volunteered to take part in this phase of the study. Participants self-reported their familiarity with virtual reality (VR) using a five-point scale. The distribution of VR experience was as Never: 20% (n = 6)Rarely: 56% (n = 17)Monthly: 10% (n = 3)Weekly: 10% (n = 3)Daily: 3% (n = 1)
The preference data from all 30 participants was pooled to assess the preference between the two audio configurations. In this data set, 20 of 30 (66.7%) participants preferred the Loudspeaker Array, while 10 (33.3%) preferred the HMD Binaural Audio.
A two-tailed Binomial Test was conducted against the null hypothesis of equal preference (50%/50%). The test revealed that the preference for the Loudspeaker Array approached, but did not reach, the standard threshold for statistical significance, p=0.099.
The 95% Clopper-Pearson Exact Confidence Interval for the proportion of participants who preferred the Loudspeaker Array was calculated as [0.472,0.827].
Although a clear majority (66.7%) of the participants preferred the Loudspeaker Array, the result is not statistically significant at the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$ \alpha
### Phase 3 qualitative results To provide a deeper look at the qualitative data, we analyzed the feedback of the participants on the native VR spatial audio (headset) versus a basic physical loudspeaker dispersion system. A thematic analysis of these experiential accounts within the four primary dimensions of Immersion, Realism, Sound Quality, and Spatial Accuracy revealed the user experience. #### Immersion *Headset* User reception was characterized by a distinct “near-field” intimacy. Some participants reported that the immediate proximity of the audio drivers enhanced their sense of immersion, citing a comfort level established by traditional VR usage. The low-latency responsiveness of the spatial engine to head movement was also highlighted as a key driver of environmental engagement. However, a subset of users critiqued this modality for feeling "closed in," noting a significant lack of perceived expansive space. *Loudspeakers* This system was consistently rated as more immersive, offering a superior sense of scale and presence that encouraged environmental exploration. Participants described a relaxing, panoramic “wide space” that felt more authentic to natural listening. Interestingly, while the headset audio felt “part of the environment”, the loudspeaker system was occasionally perceived as “detached” from the immediate visual scene, suggesting that integrated headset audio may offer a more seamless, albeit tighter, sensory fusion. #### Realism *Headset* The perceived realism of the internal system was frequently questioned. Users noted a deficiency in depth and fine-grained detail. Feedback described the audio as “fake” or “not clean”, lacking the natural resonance expected from a physical environment. *Loudspeakers* In contrast, the physical dispersion system was lauded for its high degree of naturalness. Participants reported an experience that sounded ‘less processed” and more aligned with the organic physics of real-world hearing. The ability to highlight directional nuances contributed significantly to this increased sense of environmental authenticity. #### Sound quality *Headset* Fidelity was a primary pain point for internal drivers. Participants frequently characterized the output as "subpar," "muffled," or "tight," some noting frustrating overlap in frequency ranges. A recurring critique was the sensation of the audio "hitting the ear" directly, which led to perceptions of volume fatigue even at standard levels. *Loudspeakers* This configuration was praised for its crystalline clarity and perceived depth. While the higher quality was evident, some participants noted that the inherent echo of the physical room became more prominent, which occasionally introduced a layer of perceptual confusion. #### Spatial accuracy and localization *Headset* Results with respect to localization were polarized. While many appreciated the precise centering of audio relative to their ears and the fluid response to rotational tracking, others felt the system failed to translate a true sense of 3D volume, describing it as sound "just happening" without a clear spatial context. *Loudspeakers* The physical system excelled in rendering natural distance and object placement. Users found the "proximity effect"—the ability to sense how close or far a virtual object was—to be significantly more legible and clear compared to the headset. ## Discussion Table 1Thematic analysis headset (HMD) vs. loudspeaker array across phase 1, 2, and 3ThemePhaseHeadset (HMD binaural audio)Loudspeaker arrayImmersion1Proximal audio is immersive but can feel “closed in.”Greater presence, scale, and “wide space.”2Good relation to movement; “enclosed space” feeling“Bigger space”; felt like being “actually in the scene.”3Near-field intimacy; lack of expansive perceived spacePanoramic scale; occasionally felt “detached” from visualsRealism1Lacks depth and detail; can feel “fake” or less naturalPraised for naturalness; “less processed” experience2Clearly lower quality and less realistic than speakersRobust and deep; depth felt significantly more realistic3Deficient in fine-grained detail; perceived as “not clean.”High naturalness; aligns with organic hearing physicsSound quality1Subpar/muffled; sound felt like it “directly hit the ear.”High clarity and depth; room echo can be prominent2Described as “tight,” “muffled,” or “flat.”Consistently clearer, “vastly cleaner,” and robust3Fidelity pain point; volume fatigue and freq. overlapCrystalline clarity; room echo occasionally confusingSpatial accuracy1Clear localization; responsive to rotational trackingAccurate sense of distance; effective proximity effect2Followable direction but can feel “forced.”Accurate distance/localization; echo enhances realism3Precise centering; fails to translate true 3D volumeExcels in natural distance and object placement legibility The synthesis of qualitative feedback across three phases, combined with quantitative preference data, reveals a consistent user lean toward physical loudspeaker arrays for spatial audio, although the strength of this preference varied between experimental conditions. In Phase 2, we observed a statistically significant preference for the loudspeaker array (73.3%, p=0.0063). Although Phase 3 also showed a majority preference for entry level spatial audio loudspeakers (66.7%), the result moved to a suggestive preference that did not reach statistical significance (p=0.099). As outlined in Table 1, thematic analysis clarifies these trends by highlighting a fundamental tension between the fidelity and scale offered by loudspeakers and the sensory fusion and intimacy provided by the headset. ### Fidelity as the primary preference driver A recurring theme across all phases was the perceived deficiency in the headset’s audio quality. Participants consistently described internal drivers as “subpar”, “muffled”, “tight”, or “lower quality”. This led to “volume fatigue” and a sense that the audio was “fake” or “processed”. In contrast, the loudspeaker system was praised as being “robust”, “crystalline”, and “natural”, aligning more closely with the physics of real-world hearing. This superior sound quality was likely the main driver for the significant preference observed in Phase 2. ### Scale vs. the immersion paradox Qualitative data reveal two different types of immersion that explain the split in participant preference. Presence and Scale: Loudspeakers consistently offered a “bigger space” and a “panoramic” wide field that created a greater sense of scale. This encouraged environmental exploration and was described as "just like being real". Near-Field Intimacy: Despite the preference for loudspeakers, a consistent minority (26.7% in Phase 2 and 33.3% in Phase 3) preferred the headset. Qualitatively, these users valued the “near-field intimacy” and the fact that the audio felt like a “part of the environment”. The shift from a significant result in Phase 2 to a non-significant trend in Phase 3 may be attributed to the perception that while the speakers offered better scale, they were occasionally felt to be "detached from the scene" compared to the seamless integration of the headset audio. ### Spatial accuracy and the proximity effect Localization experiences were polarized across the study. The headset was frequently praised for its fluid responsiveness to head tracking and “low-latency” engagement. However, it struggled to translate 3D volume, with sound often described as “just happening” without space. In contrast, the loudspeaker system excelled at rendering the “proximity effect” and natural distance. The ability to accurately sense the distance of virtual objects—such as the “birds flying over the top” —provided a level of spatial legibility that native HMD drivers could not match, reinforcing the majority preference for the physical array ## Conclusion The results of our study have provided valuable insight into participants’ experiences with both the in-built VR spatial sound system and the physical dispersion loudspeaker Dolby Atmos system. Although the built-in system effectively conveyed basic spatial information and contributed to a sense of presence, particularly in dynamic sound localization, it was frequently criticized for lower audio quality, inconsistent volume levels, and less natural sound reproduction. Participants generally perceived the physical space audio dispersion loudspeaker system as significantly more immersive and realistic, praising its superior sound quality, spatial clarity, and ability to create a strong sense of presence, especially in larger virtual environments. This system improved the integration of auditory and visual signals, enhancing the overall experience. These results demonstrate that loudspeaker dispersion systems greatly enhance the VR experience. There are still potential limitations of the non-portable nature of loudspeaker systems in their applicability to different VR use cases, but this brings into discussion the use of loudspeakers for other XR and augmented reality (AR) hybrid spatial audio scenarios. In the future, when designing immersive and perceptually fulfilling experiences, tools can be developed to interface VR or AR software with available loudspeaker technology to enhance the user experience in a wide range of consumer and industrial applications.