The Memory Palace of Lost Libraries

Imagine walking through a dark gallery, each room containing a single, softly glowing box. You approach, and as your shadow falls across it, a slow, whispered voice begins, not from speakers but from the box itself—a blend of recorded paper, distant footsteps, and a narrator recalling a lost library. The Memory Palace of Lost Libraries is a series of 9 dioramas, each a distinct volume in a larger story about impermanence and collective forgetting. Each diorama is a handmade miniature of a real library destroyed by fire, war, or neglect: the Royal Library of Alexandria, the House of Wisdom in Baghdad, the Sarajevo National Library, the Bibliothèque Municipale of Chartres, the Library of the University of Córdoba, the Prussian State Library in 1945, the National Library of Bosnia, the Jaffna Public Library in Sri Lanka, and the Museum of Innocence's archive in Istanbul (the last imagined). The dioramas are not realistic replicas but poetic reconstructions—rooms with tilted floors, books that grow from the walls like mushrooms, and tiny, unfocused figures. Each box contains a depth of 18 inches, with layered planes of laser-cut birch, translucent acrylic, and hand-painted textures. Hidden photoresistors and capacitive touch sensors trigger a low-fidelity audio piece (3-5 minutes) that murmurs through a hidden bone-conduction transducer embedded in the box's base, so the sound seems to come from inside the viewer's own skull. The entire project is designed for a single audience member per box, emphasizing intimate, solitary encounters. As a solo creator, every element—from soldering circuits to painting 1:100 scale book spines—is made by my hands. There is no team, no grant, no collective. The work is a meditation on what it means to hold memory alone.

Most memorials to lost libraries are textual or digital—catalogs of what was. But the sensory experience of a library—the smell, the scale, the sound of pages—is impossible to archive. There are no physical, poetic spaces that let a person privately mourn these losses. Existing museum installations are either too literal or too abstract. This project scratches the itch of tactile memory: it puts the viewer inside a dream of a library, not a replica of one.

• Complete 9 dioramas, each with unique architectural design and audio narrative.
• Develop a reliable, low-maintenance sensor system (photoresistor + capacitive touch) that triggers audio without fail.
• Each diorama must have at least 3 depth layers, with miniature books, shelves, and architectural details crafted at 1:100 scale.
• Produce original audio (ambient recordings, narration, field recordings of paper rustling and distant chatter) for each diorama, under 5 minutes.
• Exhibit the series in a gallery that allows for a single visitor path through the space, with no more than one viewer per box at a time.
• Document the process thoroughly (photos, videos, build notes) as a standalone archive.

• Not a digital simulation or VR experience. No screens.
• Not interactive beyond presence and touch. No buttons, no choices.
• Not historically exhaustive or academic in accuracy. Poetic license is paramount.
• Not a collaborative project. All making, writing, recording, and building is done solo.
• Not a performance. No live elements.
• Not portable for touring unless each diorama can be crated individually.

• Materials: birch plywood (3mm, 6mm), acrylic sheets (clear, frosted), archival book cloth, acrylic paint (matte, metallic), basswood for miniature furniture, Japanese paper for book covers, wire, solder, heat-shrink tubing.
• Electronics: Arduino Nano, photoresistors (GL5528), capacitive touch modules (TTP223), bone-conduction transducers (Dayton Audio DAEX25), low-power audio amplifier (MAX98306), microSD card module, small LiPo battery (500mAh) with charging circuit.
• Tools: laser cutter (Epilog or equivalent), X-acto knives, tweezers, magnifying lamp, soldering station, Dremel, fine paintbrushes (size 0-2), airbrush for gradients.
• Audio recording: Zoom H6 with binaural mics, contact microphones, field recording gear.
• Software: Audacity for audio editing, Fusion 360 for box design, Arduino IDE, LightBurn for laser cutting.

Each diorama is a standalone box measuring 24"W x 18"D x 14"H, with a viewing window on one face. The interior is divided into three depth planes: foreground (frame edge, <2" from viewer), midground (4-8" deep), and background (10-16" deep). The planes are cut from birch plywood or acrylic, painted and textured, then mounted on slots. Books are made individually: 1:100 scale using card stock, cloth spines, and printed covers. The electronics are housed in a concealed compartment beneath the floor, accessible via a magnetic panel. The bone-conduction transducer is glued to the inside of the top wall, so the sound resonates through the entire structure. The photoresistor sits near the viewing window; when ambient light drops (viewer blocks light), it triggers a pin on the Arduino. The capacitive touch sensor is a copper tape patch hidden under a 'welcome' text on the base. Both sensors must be triggered to start audio, preventing false starts from general gallery light changes. The Arduino reads the SD card and plays a single audio file per diorama through the transducer. Audio is stored as 16-bit 22kHz mono WAV to minimize memory usage. The system auto-powers off after 2 minutes of inactivity to save battery. The diorama's lighting comes from a single LED strip (warm white, 3000K) behind the background plane, creating a soft backglow.

• Bone-conduction transducers may not produce sufficient volume in a gallery, requiring redesign.
• Escalating time per diorama—each could take 2-3 months of solo work, leading to burnout or unfinished series.
• Sensor reliability: photoresistors may be tripped by gallery lighting changes or moving shadows not from a viewer.
• Transport or installation damage to delicate miniature elements.
• Battery life insufficient for a full exhibition day; need to design for easy swapping or external power.

• Can bone-conduction audio sound 'inside the head' for all viewers, or is it too dependent on skull shape?
• Should the audio include ambient sound from the original library location, or be entirely composed of abstract sounds?
• How to handle the inevitable degradation of the dioramas over time (paper fading, glue failing)—should there be a maintenance protocol for the gallery?
• Is 9 the right number? Too few might feel incomplete; too many might dilute impact. Perhaps 7 (one per continent) is more poetic?

I had the sketches, the electronics prototyped, and three miniatures half-built. But then I moved cities for a teaching job, and the workshop space vanished. The boxes sat in storage. I kept buying supplies, but without the original studio, the project lost momentum. Eventually, the idea felt like a past self's dream—still beautiful, but not urgent enough to rebuild from scratch. It remains a complete plan, waiting for a room of its own.

The audio narratives are the emotional core. Each should center on a single 'inhabitant' of the library—a librarian, a reader, a book—speaking in first person about the moment of destruction. The narrator is always alone. I want to record these in one take, in a room with books and a mic, to capture the acoustic loneliness. The project's title is a pun: 'Memory Palace' as mnemonic device, but also the palace as an actual dioramic space.