Forever Frequencies

As a traveling installation made to be understood and displayed to a variety of different audiences and contexts, “Forever Frequencies” lives through the materiality of the timeless music box. By evoking the latter, the idea is to channel the spirit of early childhood as a moment of both wonder and simplicity in which empty spaces begin to fill with music and memories.

The main interaction of the installation takes place when the visitor goes up to the music box and answers a simple questionnaire about a memory they have about music: where did that memory take place? When? Who was it with? Then, the narrative shifts to the future - prompting the user to imagine a future memory, a liminal space in time where the past reflects itself in the future.

Taking in the user’s answers, the installation then computes a melody, using each answer to determine and shape the different musical features of the tune (tempo, range, key…) This output is the user’s personal, one-of-a-kind “forever frequency” - their own melody based on this future memory they’d like to relive. With each visitor creating their own melody, this edifies the overall installation into a kind of joint symphony of sound and reminiscence. 

On a more technical note, the idea of the piece was to use a Murobox (MIDI-controlled mechanical music box) to play the melodies that embody the visitor's chosen music-related memory. The framework here is quite simple: a Gen-AI creating variations and tunes on a music box. At the same time, we always aim to remain critical and transparent in our decisions about the tools we use and their impact, as well as to go beyond the current randomness of AI and design a system to gain a higher control over the outcome. For instance, the artwork runs on a local LLM server, which is only available at the location of the installation - a deliberate choice we made in order to minimise our ecological footprint.
The “composition” pathway, which we fully crafted, also contributes to the artistry and intentionality of the installation - as it goes through several steps of decoding and “translating”. The user’s answers, in turn, influence the musical characteristics, which serve as “parameters” to generate a melody using the ABC musical notation, that will then be converted into a MIDI file. Once the melody is created, a second LLM model then analyses the generated song and the key concepts, to output a text that explains and contextualizes the musical memory, describing its emotional significance to the user.

In addition to the main piece where each visitor could obtain their own "forever frequency", we also collaborated with 6 artists, with 4 of them having their own music box pillar inside the exhibition. For each artist, we also asked them about a specific music-related memory - that they described in their own words, and that generated their own forever frequencies, too, for everyone to listen to and read. The artists and experts we worked with came from different fields - Kengo Kuma, an architect; Maria Arnal, an artist and composer; Suzanne Ciani, a musician and sound designer, and Max Cooper, a music producer, Richie Hawtin, an electronic musician and DJ; Daito Manabe, a media artist; Elian Chali, a muralist and activist; Ada Parellada, a chef; and Kevin Rhodes, a conductor. 

Originally, the piece was commissioned as a reworking of our 2016 “Timekeeper” installation, which we built in collaboration with the data analysis team at Spotify, for the 2016 Sónar +D. It took in the user’s Spotify musical preferences (at a time before the Spotify Wrapped!) and chose a personally curated song for them and sent it to a future date of their choice. By reworking an installation that was almost a decade old, we made sure to place our technological exploration at the center of the process - how could we integrate new advances in AI, while making choices motivated by “meaningful” artistic needs, keeping this critical perspective that is crucial to our work? We tested a variety of different tools, technologies and options, imagining the piece under all kinds of aspects and materialities, before it reached its final form. 

The installation was on display from May to August 2025 in London, and then will continue to travel around the world after a display in Japan at the Museum of Narrative, with, every time, relying on physical on-site systems. This choice, guided by a concern for environmental impact, exemplifies our desire to use AI in a transparent and conscious way, but also reclaim the purely human, analog aspect of our memories, shared experiences, and love for music.