Immersive Soundscapes with Networked Vibrating Panels: Bridging Place and Space

Immersive Soundscapes with Networked Vibrating Panels: Bridging Place and Space Mathieu Barthet1[0000000298691668],JérémyPerrouin 2,EmilyStifter 3, Vincent Debut4,5[0000000320110656], and Christine Esclapez[0000000158489409]1 1Aix-Marseille University CNRS PRISM, Marseille, France {mathieu.barthet,christine.esclapez}@univ-amu.fr 2Independent artist [email protected] 3Aix-Marseille University, Marseille, France [email protected] 4Polytechnic Institute of Castelo Branco, Castelo Branco, Portugal 5NOVA University Lisbon, Lisbon, Portugal Abstract. This paper presents the “Sound, Music and Sciences" (SMS) Lisbon workshop, a research-creation case study aimed at developing student skills in interdisciplinary practice. The project successfully engaged students in collaborative multi-channel soundscape composition, blending theoretical knowledge with practical application. The work culminated in a sound installation presented as part of the NOVA Contemporary Music Meeting (NCMM 2025). The system was developed using frugal technologies (networked Raspberry Pis and a 16-channel vibrating panel system) specifically to democratize access to spatial practice for students. This setup proved to be economical, lightweight, and portable, making multi-channel diffusion accessible and pedagogically effective. Its channel-based spatialization, while simpler than advanced algorithmic techniques, offered effective discrete source placement and immersion with negligible processing latency. The evaluation suggested significant audience engagement, characterized by cognitive intrigue, sensations of travel, and spatial rediscovery of Lisbon. The installation’s spatial nature fostered listener movement and social interaction. While the "unusual and hidden Lisbon" theme presented a paradox for local residents, the foreign composers found ample opportunities for exploration. Participants experienced parallel sensations of spatial displacement ("going somewhere") alongside a rediscovery of a city with which they were already familiar. Despite this, the feedback received was overwhelmingly positive, underscoring the approach’s potential to foster engagement and inspire future creative projects. Keywords: Acoustic ecology ·Soundscape composition ·Multi-channel sound installation ·Spatial audio ·Networked audio ·Frugal audio All rights remain with the authors under the Creative Commons Attribution 4.0 International License (CC BY 4.0). Proc. of the 17th Int. Symposium on Computer Music Multidisciplinary Research, London, United Kingdom, 2025 Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 149 M. Barthet et al. 1Introduction Sonic arts are evolving towards immersive experiences through advancements in spatial, networked, and distributed audio systems. Acoustic ecology, pioneered by R. Murray Schafer, has reshaped our understanding of the sonic environment as a compositional domain [19]. This practice increasingly uses spatial audio to create auditory illusions of sound originating from specific 3D locations [22]. Immersive sonic environments are now integral to virtual reality and public installations. However, sophisticated spatial audio systems often require significant investment, creating accessibility barriers for artists and institutions. This has led to interest in frugal technology solutions prioritizing simplicity, affordability, and accessibility under resource constraints, which are crucial for democratizing advanced sonic practices [18]. This paper presents the SMS Lisbon workshop, a collaborative initiative from Aix-Marseille University (AMU)’s "Sound, Music and Sciences" (SMS) research training program. The SMS project fosters research and creation skills among students in AMU’s interdisciplinary Master’s in Acoustics and Musicology and partnering universities. Organized by AMU, Universidade NOVA de Lisboa (UNL), and a sound artist, the SMS Lisbon workshop engaged students in creating soundscape compositions for a multi-channel networked sound installation within specific theme and timeframe constraints. This multi-channel system, by utilizing frugal technologies within a research-creation framework, directly addressed accessibility barriers, thereby establishing the core techno-social novelty of our work through both democratization and skill development. Both the system’s development and the installation’s reception were evaluated, providing insights into technical implementation, pedagogical effectiveness, and audience engagement. Subsequent sections detail related work, workshop methodology, the networked vibrating panel system, and an evaluation based on interviews and critical analysis. 2RelatedWork 2.1 Acoustics Ecology and Arts: Hearing the World’s Sounds Acoustic ecology, a nascent field emerging in the 1970s, investigates the human relationship with the world through auditory perception. This interdisciplinary field bridges music, sound, musicology, and visual/sound arts, facilitated by their aesthetic and historical permeability since John Cage’s pioneering work. In 1977, R. Murray Schafer, in The Tuning of the World [19], conceptualized the environment as an expansive composition of sounds. A soundscape is a combination of sounds forming or appearing within an immersive environment, encompassing natural sounds (e.g., animal vocalizations, wind, rain) and humangenerated sounds (e.g., daily occurrences, music, sound design). Acoustic ecology analyzes these soundscapes, whether naturally occurring or resulting from field recording and creative practices. Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 150 Immersive Soundscapes with Networked Vibrating Panels Interwoven with artistic disciplines, acoustic ecology involves philosophical, aesthetic, phenomenological, acoustic, semiotic, psychological, and psychoacoustic considerations. It is underpinned by an ethic [10] aimed at understanding the world’s sonic resonances and their interaction with the human body. Listening to landscapes, natural environments, and urban spaces unveils meanings that illuminate aesthetic, cultural, educational, economic, and social relationships. Recently, it evolved into sound ecosophy ([16]; [8], [3]), representing a civic initiative for environmental transformation and attentive listening to all sonic manifestations [11]. Fundamentally, it is a transdisciplinary field, integrating insights from humanities, exact sciences, and natural sciences. 2.2 Soundscape Composition Soundscape composition creates acoustic works from environmental sounds. Early works applied acoustic ecology principles to document and transform existing sound environments into structured compositions [19]. More recent efforts integrate new technologies, moving beyond fixed media. Contemporary compositions leverage tools like Ambisonics for spatialization [4], networked audio systems for diffusion [24], data sonification [17], and openly licensed audio from databases like Freesound [15]. This blurs the lines between soundscape composition and broader sound installation practices [1]. Technological advancements enable dynamic spatialization and novel presentation formats, expanding acoustic communication principles [22] and pushing audience engagement boundaries. 2.3 Frugal Technologies for Spatial and Networked Audio Frugal technologies prioritize simplicity, affordability, and accessibility. In sonic arts, they are increasingly applied to spatial and networked audio, which distributes sound in space and place to create immersive environments. The high cost of traditional spatial audio installations drives research into frugal methods, especially those leveraging networking capabilities. This includes networked microcontrollers and micro-computers; Chafe and Oshiro [6] demonstrated JackTrip’s feasibility on Raspberry Pi for cost-effective, low-latency streaming. Michon et al. [14] showcased FPGAs for frugal Wave Field Synthesis (WFS) systems, achieving a 32-speaker setup for under $800, enabling scalable and accessible spatialization over networks. Web-based frameworks also leverage ubiquitous devices for frugal networked spatial audio. Matuszewski [13] expanded soundworks for distributed multimedia web applications, where audience devices act as networked sound sources. Bevilacqua et al. [5] explored web technologies for collective interaction. Rushton et al. [18] introduced a decentralized client-server system using low-cost microcontrollers for distributed spatial audio, addressing scalability and synchronization via UDP multicast. These efforts lower the barrier to entry for spatial and networked audio, emphasizing accessible hardware and open-source paradigms. Our work focuses on a Raspberry Pi networked system using lightweight vibrating panels for accessible multi-channel diffusion in a pedagogical context. The Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 151 M. Barthet et al. novelty of our contribution lies in this techno-social application: providing a reproducible, low-cost system specifically engineered to facilitate interdisciplinary student collaboration and democratize access to multi-channel sound creation. 3SMSLisbonWorkshop:AResearch-CreationApproach 3.1 Workshop Theme and Structure The workshop, a collaboration between AMU, UNL, and sound artist Jérémy Perrouin, involved 18 students from AMU’s Master’s in Acoustics and Musicology and UNL’s Master’s in Musical Arts. Spanning six months, it began with organizers defining themes and logistics. The workshop took place from May 2-8, 2025, at UNL, focusing on creating soundscape compositions for public sound installation using networked vibrating panels. This integrated aesthetic and technological goals, aligning with university curricula and a research-creation methodology. Compositions explored themes of an unusual and hidden Lisbon, urban memory, and cultural heritage. The six-day workshop began with introductions and theoretical sessions on acoustic ecology and soundscape composition, followed by hands-on labs covering Raspberry Pi and networked audio, vibrating panel systems, and field recording. Participants formed four interdisciplinary student groups (four or five students from both AMU’s humanities/engineering streams and UNL). A day was dedicated to field recording in Lisbon and developing composition ideas. The latter half of the week included a theoretical session on vibrating panel acoustics, group mentoring, intensive group work, finalization of creative and technical aspects, and installation/testing. The workshop concluded with a sound installation at the NCMM 2025 conference held in the outdoor pergola at UNL’s Berna campus. An evaluation was conducted through interviews during this installation. Field recordings used Zoom H4 and H5 recorders with windscreens. A Discord server facilitated dissemination of presentations, lab files, and documentation. 3.2 Research-Creation Methodology The workshop and the SMS project employ a research-creation (RC) methodology, often interchangeable with practice-based research (PBR). Both define research where creative practice is integral to generating new knowledge, which can be embodied, tacit, experiential, or quantitative. The creative act functions as an investigation itself. Giacco and Gosselin [9] define RC as “a field of research rooted in academic and institutional contexts within the contemporary art world... a hybrid methodology fostering constant back-and-forth exchanges between ’artistic production and theoretical reflection’”. Both RC and PBR emphasize a reciprocal relationship where theory and practice mutually inform each other, often involving a cyclical process of creation, reflection, analysis, refinement, and subsequent creation. Bahng et al. [2] distinguish PBR’s aim to generate knowledge Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 152 Immersive Soundscapes with Networked Vibrating Panels through practice from RC’s emphasis on social and community-based engagement. While often solo, RC also develops collaboratively, prioritizing joint execution of projects, as Stévance and Lacasse [21] emphasize. In music, RC "offers a space for dialogue and exchange aimed at bridging knowledge and know-how, respecting the respective requirements of music research and musical practice" [21]. The workshop effectively utilized RC’s interdisciplinary nature and collaborative experimental project management, bridging theory and practice with a non-hierarchical convergence of knowledge. 4DesignoftheNetworkedVibratingPanelSystem 4.1 Overall Design Requirements The acoustic system design was constrained by the workshop timeline, location, and pedagogical context. It needed to be lightweight, economical, and easily transportable across countries. Quick setup in an outdoor public space was essential, and sound diffusion technologies had to be readily learnable by students within a limited timeframe, using open technologies. 4.2 Multi-channel Networked Vibrating Panel System To meet these requirements, we developed a 16-channel audio system based on vibrating panels. These panels were hardwired to Raspberry Pi micro-computers and micro-amplifiers, wirelessly connected to a client computer. PureData served as the audio engine, allowing wireless configuration, portability, and ease of integration. The panel system’s design and implementation are detailed next. Fig.1: Side acoustic panel suspended within the pergola (left) and listeners experiencing soundscapes (right) Design and Implementation of the Sound Installation The sound installation used vibrating acoustic panels, actuated by electrodynamic transducers. The objective was to create an immersive, diffuse auditory experience outdoors Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 153 M. Barthet et al. using a discreet, aesthetically mindful solution, while raising student awareness of perceptual considerations in sound spatialization. Figure 1 (left) shows a side acoustic panel suspended within the pergola. Panel Material and Acoustic Properties The acoustic panels were 5 mm thick A1-sized foamboard (carton plume). This material was chosen for its rigidity, which facilitates vibration transmission, and its low weight, which minimizes energy demands on exciters, enabling frugal audio technology. Its inherent internal damping mitigates prominent self-resonances and contributes to a relatively balanced frequency response. Fig.2: Rear view of the acoustic panel with electrodynamic transducers Fig.3: Spatial arrangement of the acoustic panels within the pergola Transducer Configuration and Sound Diffusion Each panel was actuated by two electrodynamic transducers (Dayton audio EX25VT2-4, 20W, 4 Ohms) affixed to its rear surface (see Figure 2). Transducer placement was empirically fine-tuned to optimize broad sound diffusion and prevent listeners from localizing the exact exciter positions. This configuration enhances the perception of an extended, ethereal sound source. The panels exhibit low-directivity sound radiation, fostering sonic envelopment even at close proximity. Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 154 Immersive Soundscapes with Networked Vibrating Panels Spatial Arrangement and Installation Logistics The installation’s location within UNL’s outdoor pergola (approximately 12m x 5m) facilitated unhindered listener movement. The aim was a 16-channel immersive experience with pervasive sonic presence, accommodating audience mobility while minimizing visual intrusiveness. The panels were spatially arranged to create a three-dimensional immersive field (see Figure 3). Eight panels were suspended vertically around the perimeter, and eight horizontally above the listening area, forming an upper sound layer. This spatial distribution fostered a permeable acoustic bubble where sounds could originate from any direction. Suspension System Design and Implementation The suspension system was engineered for logistical, climatic, mechanical, and aesthetic constraints. Edges were folded 90°(2 cm deep) to create a rigid frame, held by custom 3Dprinted brackets with screws and silicone glue. This prevented warping and allowed discreet perforations for slender, tensioned suspension wires. Anchored at elevated points, the wires allowed panels to visually "float," maintaining vibration freedom. The absence of visible structures enhanced the aesthetic impact, intriguing listeners about their operation (see Section 5). 4.3 Vibratory Measurements and the Physics of Radiating Panels Vibrational measurements were conducted to understand panel vibratory motion and acoustical behavior. Characterization focused on the fully coupled system (panel and two exciters as static load). Single admittance measurements were performed by impacting the panel with a hammer and measuring velocity response with a laser vibrometer (Figure 4). Excitations were applied at exciter locations, and vibratory response was measured near a panel corner. As Figure 4 shows, admittance exhibits strong modal character up to 600 Hz, remaining constant at higher frequencies where resonances are less pronounced. The exciter’s frequency range, with impedance peaks at approximately 100 Hz and 10 kHz (manufacturer data), delineates the useful frequency range for the exciter-panel coupled system. While limiting low-frequency reproduction, this favors system linearity above 100 Hz, mitigating strong filtering from low-frequency modes. Anotablecharacteristic,perceivablebyrotatingthepanel90°,isthatpanel sound radiation is dominated by dipole behavior. This results from the unbaffled configuration, allowing wave cancellation between both faces, leading to small low-frequency radiation followed by a rapid increase with frequency. Understanding sound radiation from planar structures remains difficult, depending on surface geometry, boundary conditions, baffle presence, surrounding fluid effects, and excitation form [12,23]. Common analysis approaches address (1) modal radiation (sound from a specific mode) and/or (2) overall radiation from point force excitation (many modes contributing). For modal radiation, a plate radiates by its whole surface, edges, or corners, with efficient radiation occurring when acoustic wavelengths are smaller than structural wavelengths. Surface modes are most efficient, followed by edge and corner modes. For point excitation, all three families of modes can be simultaneously excited. Low-order modes can contribute Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 155 M. Barthet et al. significantly to radiation even if excitation frequencies are far from their natural frequencies, due to modal damping [7]. Improving radiation efficiency, particularly by coupling modal vibration and radiation models [20] with optimization algorithms, is a clear area for future acoustic panel optimization. Fig.4: Experimental set-up (left) and measured admittance curves (right). 4.4 Networked Audio System Architecture and Implementation The sound installation used a distributed architecture with 8 Raspberry Pi 5 (Pis) microcomputers, each an independent audio playback unit. Pis were chosen for their compact size, processing capabilities, and audio hardware compatibility. Although Pi 5 specifications exceed basic audio playback needs, they were selected for future workshop scalability and advanced feature integration. Hardware Configuration Each Raspberry Pi 5 had an IQAudio DigiAMP+ via its 40-pin GPIO header. The DigiAMP+ incorporates a Texas Instruments TAS5756M stereo amplifier, providing 2 x 35W output, driving two of the 16 passive acoustic panels. Power was supplied by a KFD 65W 19V 3.42A adapter, connected via panel-mounted barrel connectors for outdoor stability. The DigiAMP+ supplied 5.1V at 2.5A to the Raspberry Pis 5. For durability, each Pi and DigiAMP+ were housed in a KKSB DigiAMP+ Case. A Pi5 Active Cooler was mounted to each board for optimal operating temperatures, especially during continuous outdoor operation. The array of Pis was secured to the pergola structure with cable ties. Portable batteries were impractical due to high cost and transport restrictions. Software Environment Pis ran Ubuntu with Pure Data (PD) "vanilla" (v0.552) for real-time audio. Audio files were stored on 32 GB SD cards. Pis acted as servers, remotely controlled by a MacBook client for starting/stopping audio. Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 156 Immersive Soundscapes with Networked Vibrating Panels Network and Remote Management A dedicated NETGEAR router established the wireless network for the installation. All 8 Raspberry Pi servers and the client MacBook connected to this network for seamless control and data transfer. System configuration and file management on the Pis used GUI tools (VNC Viewer, FileZilla Client) and Terminal commands (e.g., ssh, scp, rsync). iTerm2 and i2cssh (Mac OSX) or cssh (Linux) enhanced efficiency by enabling synchronized command execution across the Pi cluster. The local network enabled very low latency for control operations. Audio transmission between Pis and panels had no latency as they were hardwired. Networked latency and jitter were not issues, as the installation lacked interactive controls. Audio System Control and Asset Management Aspecificassetmanagement and routing protocol was implemented for multi-channel audio diffusion using Raspberry Pis and PD. MIDI controlled the audio playback system as a sampler: MIDI note values triggered distinct audio recordings on designated output channels, while Control Change (CC) messages managed individual audio file volumes. Each Raspberry Pi operated identically, with all necessary audio and MIDI files replicated. The system had a client ("master") controlling multiple server nodes ("local" Raspberry Pis). A PD program on the client served as the central control unit, managing master gain and triggering MIDI file playback on each networked Raspberry Pi. Individual Raspberry Pis, as server nodes, ran a "local" program for playing audio on specific output channels. These programs dynamically loaded MIDI files based on the Raspberry Pi’s hostname and assigned channel, with MIDI notes triggering corresponding audio files. Audio playback was handled by a PD sub-patch, with volume control via MIDI CC messages. Communication between master and local Pis used OSC messages for control commands. Audio processing activated automatically on startup. To manage asset complexity and ensure sufficient storage, the number of audio files per composition was defaulted to 16. Students produced 16 uncompressed WAV files (16 bits, 44.1 kHz) adhering to naming conventions, and 16 corresponding MIDI files for triggering and volume control. Each group’s project folder was copied to every Raspberry Pi. For execution, the main PD patch ran on each Raspberry Pi in non-GUI mode, and the client program on the control computer managed the system. Synchronization among Pis used Ableton Link. Networked communication within PD used OSC messages via netsend and netreceive objects with UDP for lower latency. Spatial Audio Implementation Each audio file or sound event was directly assigned to a specific acoustic panel, connected to a dedicated output channel. Spatial movement was achieved by manual panning, crossfading between fixed channels, or sequential/concurrent triggering. While 16 panels acted as discrete sound sources, this configuration enabled spatial audio through: (1) Discrete Source Placement: 16 distinct points of sound origin allowed listeners to perceive shifts in sound localization based on proximity and orientation to active panels. (2) Panning/Movement by Amplitude: Sound movement was genProc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 157