Designing a Geolocated Soundwalk Player Using DIY Electronics: An Underestimated Approach to Site-specific Sound Art and Research

Designing a Geolocated Soundwalk Player Using DIY Electronics: An Underestimated Approach to Site-specific Sound Art and Research Denez Thomas1*[0009000008582891],AntoineBeaufort 2, and Charles Verron3 1Rennes 2 University, Rennes, France [email protected] 2Ars Nomadis, Rennes, France [email protected] 3Noise Makers, Rennes, France [email protected] Abstract. This paper presents a geolocated soundwalk player developed for the artist collective Ars Nomadis (Rennes, France). Unlike most existing options, this player is not a smartphone application, but a dedicated DIY electronic device designed to be as simple and reliable as possible. It comes with an open-source web editor consisting of a map interface and an inspector to place and configure listening areas based on a set of parameters and conditions. This paper discusses both the technical aspects, including the reasons that led to the design of a dedicated player rather than the use of smartphones, and their observed consequences on musical and sound composition in public spaces after nearly a year of on-site experimentation. For art and research purposes, both the modularity of rapid prototyping techniques and the reliability required for geolocated audio works are needed. The results show that the two are possible through the use of such player, and suggest promising perspectives for site-specific sound creation. Additionally, informal user feedback indicates that the player’s design can positively affect the quality of the relation between sound and place. Keywords: Soundwalks ·Locative Audio ·DIY Electronics ·Embedded Systems ·Sense of Place ·Sound and Music Computing 1Introduction Ars Nomadis 4 is a non-profit collective of artists based in Rennes (Brittany, France) which specializes in musical and sound creation in public spaces. For many years, Ars Nomadis has been creating site-specific installations, concerts and soundwalks, systematically involving local residents in the creative process. The soundwalks guide listeners through the city for approximately an hour telling stories (real or fictional) about the place, its daily life, or its history, bringing out 4https://www.arsnomadis.eu/english Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 859 2D.Thomasetal. details of the environment that one might not have noticed otherwise. Together with the musical and sound composition, this poetic interpretation of the city while walking aims to contribute to the emergence of a unique sense of place. For previous soundwalks, Ars Nomadis used to release a single fixed-length audio file that listeners could stream or download and play on a smartphone or any portable audio player. This is probably the easiest and most robust way to compose a soundwalk, but the fixed duration imposes the same walking pace on each person. Geolocation opens a more subjective and engaging approach. For example, the listener can choose between different paths to take, or enjoy a particular place for longer. This is part of the reason why geolocated soundwalks have been adopted by both artists and researchers across diverse practices. For instance, Cloves [ 1 ]studiedtheireffectivenessforresearchandeducationin historic outdoor environments. More broadly, locative audio setups have been used as tools to explore the relationship between sound and the experience of place [ 8 ]. Harju [ 6 ] has also examined their site-specific narrative potential within the field of Audio Augmented Reality (AAR). Fig. 1. Two listeners equipped with the Ars Nomadis Player during a soundwalk. Most existing platforms for creating geolocated soundwalks, such as Echoes, 5 SonicMaps, 6 and SoundWays, 7 rely on a web editor and a mobile player. While feature-rich, these tools are not open-source, which limits their customizability for research and artistic experimentation. This limitation led to the development of a custom system: the Ars Nomadis Player (Figure 1)anditscompanionweb editor, the Ars Nomadis Editor. For reasons detailed later in this paper, the decision was made to design a dedicated open-hardware device rather than a 5https://echoes.xyz 6https://sonicmaps.xyz 7https://soundways.eu Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 860 Designing A Geolocated Soundwalk Player Using DIY Electronics 3 smartphone application. Its development is the result of collaboration between artists and researchers, 8 each of whom has their own expectations. For Ars Nomadis, reliability, ease of use, and low production costs are the three main requirements, followed by requests for custom features. For academics, it is the player’s ability, through modularity and rapid prototyping, to become a useful hardware platform for multidisciplinary research. The conclusion was that designing a player using do-it-yourself (DIY) techniques was the best way to meet all these needs. DIY practices are very common for both art [ 7 ]and research [ 4 ], but soundwalks have a rather specific and critical need for reliability. For instance, listeners could get lost or endangered by poor design decisions and software bugs [ 1 ]. Conversely, it can be hypothesized that a bug-free experience has the potential to offer listeners a deeper connection with the place, the sound, and the music. Nearly a year of on-site experimentation and dozens of public soundwalks now demonstrate the relevance of this approach, which encourages its further development. To the best of the authors’ knowledge, no comparable openhardware soundwalk player with an open-source editor is currently documented. 2 A Geolocated Audio Player The player is a 10 x 7 x 3 cm 3D printed (PETG) enclosure with embedded electronics built around the Teensy 4.1 microcontroller development system 9 from PJRC, and a few external features which are listed in Figure 2. It is designed to be worn across the chest like a belt bag with an additional pocket containing the GNSS antenna. Listening is done via headphones thanks to the mini jack connector on the front, which also makes it possible to connect a splitter if two listeners wish to use a single player. The device is powered on using a small switch that is intentionally difficult to operate, preventing listeners from turning it offby mistake. Two push buttons are used for volume control, and a third allows the playback of sounds freely assigned by the author. The latter is mainly used to trigger the first sound of the walk, but also as a "help" button in case the user has difficulty finding their way. The decision to design a minimal interface—with only a few buttons and no screen—was informed by prior research showing that reducing visual interaction with the device helps users focus on the soundwalk [11]. 2.1 Firmware The code running on the player is written in C++ using the PlatformIO embedded software development ecosystem. 10 The firmware is open-source (under the MIT 8 Electronic design (Noise Makers and Denez Thomas, PoC by Gabriel Chiapello), software engineering (Denez Thomas, PoC by Gabriel Chiapello), design and 3D printing of the case (Denez Thomas), assembly of the units (Denez Thomas and Marin Esnault), design and manufacture of the shoulder straps (Vanille Hurel), feedback on specifications (Frédéric Bimbot), testing and user feedback (Ars Nomadis). 9https://www.pjrc.com/store/teensy41.html 10 https://platformio.org Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 861 4D.Thomasetal. 1. Headphone Jack Connector 2. Micro-USB Connector 3. Shoulder Strap Slot 4. External GNSS Antenna Connector 5. Volume Up Button 6. Volume Down Button 7. Status LED 8. Help Button 9. Concealed On/OffSwitch 10. Battery State of Charge Indicator 1 10 2 3 45 6 7 8 9 Fig. 2. The external components of the soundwalk player. license) and available on GitLab. 11 Its repository also contains documentation and hardware files, including the models necessary for 3D printing. Soundwalk Manager. The Soundwalk Manager is the main service of the player’s firmware. Every second, it receives an updated set of coordinates and compares it to the soundwalk data. The latter is a parsed JSON file that was generated from the web editor and transferred to the player’s storage. It contains all the zone locations, radius, and the parameters of the associated audio files. Based on zone entries and exits or button presses, the Soundwalk Manager can search for sounds that meet the conditions to be played or stopped. Audio Engine. The soundwalk audio playback uses PJRC’s Audio Library 12 which supports streaming 16 bits, 44.1 kHz files from a microSD card. The firmware’s audio graph consists of four stereo players that can run simultaneously. When a new sound needs to be played, the Audio Engine searches for an available player and starts playback with a possible Fade In. 2.2 Internal Components The internal components of the player (Figure 3)arebuiltaroundtheTeensy 4.1, a development board based on an ARM Cortex-M4 microcontroller. The current configuration includes a GNSS module, 13 an audio interface, 14 acharging controller, 15 and a battery. Together, these few components represent the minimal set required to build a fully functional soundwalk player: a battery-powered, 11 https://gitlab.com/ArsNomadis/arsnomadis-player 12 https://www.pjrc.com/teensy/td_libs_Audio.html 13 The NEO-6 geolocation chip, while popular among DIY enthusiasts for its affordability, has reached End-of-Life and the use of a newer GNSS module is planned. See: https://www.u-blox.com/en/product/neo-6-series 14 PJRC’s Audio Adaptor: https://www.pjrc.com/store/teensy3_audio.html 15 Adafruit’s PowerBoost 500C: https://www.adafruit.com/product/1944 Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 862 Designing A Geolocated Soundwalk Player Using DIY Electronics 5 location-aware audio DSP system. Based on measurements, the device achieves about 8.5 hours of continuous operation. Longer runtimes can be achieved by using a higher-capacity battery, but the included 2600 mAh unit has proven sufficient for a full day of events involving multiple consecutive soundwalks with no recharging opportunity. The design also includes connectors for two optional modules: a Wi-Fi module, for instance to enable wireless updates of soundwalk data, and a head-tracking unit (IMU) that can be mounted on the listener’s headphones for HRTF-based sound processing. 1. GNSS Module 2. Audio Adaptor Board 3. Teensy 4.1 Microcontroller 4. IMU Connector (Opt.) 5. Main Board 6. MicroSD Card 7. 18650 Li-ion Battery 8. Charger Board 9. Control Board 10. Wi-Fi Module Connector 3 9 8 7 6 5 2 1 10 4 Fig. 3. The internal components of the geolocated soundwalk player. The player can be assembled by DIY enthusiasts or technically inclined artists and researchers. While not optimized for mass production, its modular, openhardware design makes it particularly valuable to these communities, encouraging the creation of customized versions by adapting both hardware and firmware to their specific needs. 2.3 Comparison with Using a Smartphone Application Since most smartphones have all the features needed for a soundwalk (namely geolocation and audio playback), they were the first option considered for the collective’s projects, since it would have allowed everyone who owns one to use it. However, it was decided not to use the listeners’ own devices for several reasons. Firstly, a wide variety of smartphones would have been encountered, which would complicate software development. The management of geolocation permissions, for example, varies between operating systems and even between versions of the same one. Moreover, the battery capacity of some smartphones might not be enough to last the entire soundwalk. More generally, using the same device for all listeners limits the risk of technical issues related to the specific characteristics of each smartphone. Choosing a single smartphone model to distribute to listeners Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 863 6D.Thomasetal. did not seem like a better alternative due to the difficulty of finding one that met all the needs of a soundwalk at a reasonable price. In terms of geolocation performance, the Ars Nomadis Player achieves about 2-5 m CEP50 horizontal accuracy depending on sky visibility, a level comparable to that of a recent smartphone. 16 Providing listeners with the same device used during the soundwalk’s composition ensures that zone placement and sizing are designed with known performance in mind. While the current accuracy is sufficient for Ars Nomadis’ soundwalks, upgrading to a more recent GNSS module would offer higher precision while preserving this predictability. The previously cited study by Cloves [ 1 ]evaluatedfifteensoftwareplatforms using criteria similar to those required for Ars Nomadis’ geolocated soundwalks. As all scored platforms use a mobile application as player, some of the criteria are related to smartphone-specific issues: "Can the app be used without looking at the phone?", "Can the app function while focus is on another app?", or "Does the app respond intelligently to incoming phone calls?". These requirements highlight common reliability concerns when working with smartphones, particularly related to the fact that these devices have many other functions than being soundwalk players. They can run multiple concurrent applications, go into sleep mode, receive calls and notifications, etc. While most of these smartphone-specific issues can be addressed either by application developers or by asking listeners to change device settings, the hypothesis was to consider it more advantageous to develop a dedicated system. By focusing on modularity and the selection of well-known and replaceable components, the aim was to ensure better repairability and better resistance to obsolescence [3]. 3ADedicatedWebEditor The editor can be seen as a graphical interface for generating the JSON file that will be interpreted by the player. Its map interface makes it easy to place, size and visualize listening areas, while its inspector (on the right in Figure 4)isused to configure sound parameters. Like the player’s firmware, the editor’s source code is available under the MIT license, and its repository is hosted on GitLab. 17 The running web application is accessible via the Ars Nomadis website.18 Since the editor is completely separated from the player, two options were considered for its practical implementation: a standalone application (potentially based on a game engine) or a web-based interface written in JavaScript. The latter approach was chosen for two reasons. First, web applications are straightforward 16 Static horizontal positioning accuracy was assessed using geodetic reference markers ( n = 1000 fixes per condition). Under open-sky conditions, the player reached 1.93 m CEP50 and 2.36 m CEP95, compared to 1.91 m CEP50 and 1.99 m CEP95 for the smartphone (Asus Zenfone 10). In a partially obstructed urban environment, the player reached 4.48 m CEP50 and 12.92 m CEP95, compared to 8.93 m CEP50 and 11.24 m CEP95 for the smartphone. 17 https://gitlab.com/ArsNomadis/arsnomadis-editor 18 https://editor.arsnomadis.eu Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 864 Designing A Geolocated Soundwalk Player Using DIY Electronics 7 Fig. 4. Screenshot of the Ars Nomadis Editor, a web-based tool for creating soundwalks by placing zones on a map and configuring their audio and interaction parameters. to deploy to end users, as they require no installation and can be accessed via a standard web browser, providing broad cross-platform compatibility. Second, they benefit from powerful open-source geomatics APIs, which facilitate the development of custom mapping features [5]. 3.1 Soundwalk Composition The basic principle of composing a soundwalk with the editor is that within each zone placed on the map, one or more sound files can be added with their own configuration (see Table 1). It is the combination of sounds with different parameters and conditions that makes it possible to compose dynamic and nonlinear walks. Apart from Fade In and Fade Out, the editor does not include any sound processing parameters. This design choice was made to avoid duplicating the features of specialized tools that artists and sound engineers already use, such as digital audio workstations, for creating the narrative and musical content of the soundwalk. 19 However, the Teensy’s official audio library has many features, and more are constantly being developed by the community. A particularly powerful example is the ability to use the Faust programming language to further extend the Teensy’s audio applications [ 9 ]. All this makes the player a very useful hardware platform for experimenting with location-based sound 19 This approach is used by major soundwalk editors such as Echoes Creator. See: https://docs.echoes.xyz/docs/creator/elements Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 865 8D.Thomasetal. Table 1. The main parameters associated to an audio file in the web editor. Name Description Filename Name of the audio file. Play Event Event triggering the playback. Stop Event Event that will stop the playback. Fade In Fade in time in seconds. Fade Out Fade out time in seconds. Loop Whether the file should loop or not. Ignore Other Zones Ignores the entering of other zones during playback. Prevent Button Use Prevents the use of the Help button during playback. Zone Conditions Enables the selection of zones (including this one) that must have either already been visited or never been visited for the file to play. synthesis and processing. For example, the partner company Noise Makers 20 was able to implement its own HRTFs and experiment with spatial audio processing. Events. The most important settings in the editor are probably the selection of a Play Event and Stop Event. A sound can start either by entering a zone, leaving it, or by pressing a button. It can also be configured to start at the end of another sound, which is often used as way to automatically start a music loop after a vocal indication. Stop Events include leaving the zone, 21 entering a new zone (any or a specific one), or "None" (when the audio should play entirely). Zone Conditions. During the soundwalk, the player keeps track of "Already Visited" and "Never Visited" zones. This allows a sound to be played based on Zone Conditions, which can be combined in an "all or none" (boolean AND) manner. For example, in Prenez soin d’elles, Zone Conditions were used so that when entering the same zone, a different sound is played depending on where the listener is coming from. The author can even place soundless zones that are there only to identify which path has been taken. These conditions are also used for reliability purposes: accidental retriggering of the sound when the listener is at the edge of the zone, for instance, can be easily avoided by adding a "Never Visited" condition. 3.2 Soundwalk Simulation for Debugging As soundwalks become more complex, with many zones and conditions, testing and debugging becomes increasingly important. While in situ testing is essential, 20 https://www.noisemakers.fr 21 For reliability reasons, it is not always recommended to set the Stop Event to "Exit". When the listener is near the edge of the zone, inaccuracy of the geolocation can cause the playback to accidentally stop. The impact of this problem could be reduced with ahysteresissysteminwhichtheexitperimeterislargerthantheentryperimeter. Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 866 Designing A Geolocated Soundwalk Player Using DIY Electronics 9 using a geolocation simulator can prevent many easily avoidable issues, such as incorrectly configured sound settings. This is possible by connecting the player to the editor using a USB cable. When the simulator is enabled, clicking on the editor’s map interface will send spoofed geographic coordinates (as fake NMEA sentences) to the player via serial communication (using the Web Serial API 22 ). The player then stops listening to the GNSS module, but performs as if it were receiving real geolocation data: the author can listen to the soundwalk using the headphone output, and press buttons. Logs are also sent to the navigator’s JavaScript console for more advanced debugging and monitoring. 4CaseStudy:Prenez soin d’elles (2024) Prenez soin d’elles 23 is listened to at events where 10 to 20 listeners gather at a specific location in the park of the Centre Hospitalier Guillaume Régnier (CHGR), an institution specializing in mental health. They are equipped with geolocated audio players, and after a short introduction, they press a button to start the soundwalk. The story begins in front of a cultural center, where it is said that two patients are missing, despite being registered for a creative activity. The listeners set out to find them and meet fictional caregivers and hospital staff along the way. They tell them about the place and talk about the difficulties they face on a daily basis. The fiction is inspired by interviews conducted at the hospital by Ars Nomadis since 2021 which reflect the current mental health crisis in France [ 2 ]. Despite the difficulties and growing needs, the characters defend the public hospital and imagine possible futures. 4.1 Composition Principle In Prenez soin d’elles,soundsaredividedintothreecategories:portraits,music loops and indications. Portraits, which combine voice and music, constitute the bulk of the narrative content through the encounter of a character in a given location. The music loops are then automatically triggered to accompany the walk to the next portrait. Finally, indications are directions suggested by a hospital intern who virtually accompanies the listener throughout the entire soundwalk. 24 His frequent indications, unlike portraits, can play over the music loop. A diagram of the interaction between the sound categories is shown in Figure 5. 25 Note how the music loop is triggered a few seconds before the end of the portrait and stops shortly after the next one begins. When a sound is scheduled to play after another (Play Event: After {sound}), its Fade In is applied to ensure a smooth 22 https://developer.mozilla.org/en-US/docs/Web/API/Web_Serial_API 23 https://www.arsnomadis.eu/projets/les-promenades-sonores/prenez-soin-delles 24 All indications were recorded in situ to better match the environment, both acoustically and in terms of what the actor says. This is based on the assumption that a large amount of on-site improvisation plays an important role in perceived authenticity. 25 Figure 5is a conceptual representation and does not depict the actual authoring interface. For a view of the editor interface, see Figure 4. Proc. of the 17th International Symposium on CMMR, London, UK, Nov. 3-7, 2025 867