Three times gather on one stage ── Reflections on the 5/22 performance of "DEEP FAKE: Ghost in the Machine"
The special lecture and performance of "DEEP FAKE: Ghost in the Machine" recently concluded successfully at the Showa University of Music's Yuli Hall in Kawasaki. I would like to express my heartfelt gratitude once again to everyone who made the trip to the venue, my co-creators Rosie Kay and K2CO Ltd, the three dancers—Ken Nakagawa, Yuko Shimoji, and Saaya Takaoka—our moderator Mayumi Taguchi, interpreter Fumi Matsuoka, as well as Professor Yoshio from Showa University of Music, the composition department, and our sponsors, Arts Council England and The Great Britain Sasakawa Foundation.
In the conclusion of my previous note article, I wrote the following:
"The dancer's movements, the sound of the piano, and the sound and light returned by the AI. Each marks time with a different rhythm. On stage, those three times gather in the same place. We have yet to decide whether to call it a 'dialogue' or a 'symbiosis'."
Now that the performance is over, I would like to unravel these questions of "three times" and "dialogue or symbiosis" once more, to organize what happened on 5/22 and what we will carry forward toward the premiere at the Birmingham REP in the UK in October 2026.
Three times gather on stage
During "Distant Echo," I was creating a work where three time axes—"the time of the body," "the time of technology," and "the time of hearing"—coexisted in the same space while remaining out of sync. As the next step, DEEP FAKE is an attempt to gather three times—the human body (the dancer), the piano (myself), and the AI (another person)—onto a single stage.
The dancer lives in the time of the body. A singular time where the weight of stepping on the floor and that weight itself beckon the next movement. I live in the time of the piano. A time like breath that opens up between placing one note and the next. The AI possesses a statistical and non-human time, referencing vast amounts of past learning data in an instant and responding on a scale different from either of ours.
I still cannot decide whether these three times become a "dialogue" or a "symbiosis." But at the very least, I believe that the act of them being present in the same place on stage is a practice necessary between current technology and ourselves.
The composer Mr. Takayama, who watched the performance, wrote in a thoughtful critique that "Nakamura's delicate touch and improvisational technique 'pre-conceive' a composition fraught with tension between piano and human." I feel as though he has articulated from the outside the core part that I myself had not been able to put into words well, and I would like to cherish this as the axis that concludes this interim report on the work.
Part 1 ── Reimagining the AI of More Friendly Times
In the first part, what I wanted to establish on stage was to properly redraw AI not as an "enemy" or a "convenient tool," but as a "relationship from a time when it was friendly."
The sense of distance from the dawn of the internet, where you would wander into someone's handwritten HTML page and quietly touch the voice of the person on the other side. When the AI responded to the very first note of the grand piano in Set 1, I wanted to share that sense of distance, like a memory, with the audience. I placed "ambient" music, as proposed by Brian Eno—music heard as an "environment" in daily life—at the base. The AI was neither an adversary nor a subordinate, but stood alongside the dancers and me as another presence within that environment.
I will open up the implementation side just a little. The piano I play is captured in real-time by the PC's AI, and piano internal techniques, including my new album scheduled for release next month was learned by Somax2, which returns sounds as if responding to my performance. Similarly, RAVE, which also learned from that album, generates effects on the resonance on the spot. Furthermore, as the performance progressed, the dancers' body movements were captured, and sounds from an AI that learned my real-time piano playing were returned for each body part was the structure.
And on stage, as a consequence of these systems, another existence—a **ghost (daemon)**—begins to move. The ghost starts by mimicking the dancers' movements, gradually transforming its appearance into something like a specter—or in Western terms, something otherworldly—and finally begins to dance Rosie's own dance. My intention in the first part to stand AI up as "another person" may have been most directly manifested in the very gestures of this ghost.
However, to be honest, due to time constraints on the day, I was unable to provide a space to explain to the audience what this system was doing, and I could not incorporate scenes into the performance itself where the mechanism was intuitively "visible." I feel there is still a bridge to be built between the state where technology is moving complexly behind the scenes and how that reaches the audience's bodies as a tangible experience.
Part 2 ── Being Tossed About, and Reclaiming Agency in Sound
The theme of the second part was how humans behave after new technology appears. Even when new technology appears, humans themselves cannot easily grow, and we continue to re-grasp classical things over and over again—just as we have been doing ever since the Renaissance.
The deconstructed club, noise, and glitch music placed in the first half are the sounds of that "time of being tossed about." A state where the human body is swept up, unable to keep up with the unpredictable rhythms output by the AI. The free jazz in the second half is a time for humans to recover the state of "being able to listen as a work of art" to the scattered sounds once again, through their own learning—the act of playing the piano—after experiencing that turmoil. I chose the breathing sensation of free jazz not as a return to the classics, but as a question of stride when walking alongside technology.
Here, too, I would like to write frankly about what happened between the design and the actual performance. What I had originally incorporated in Set 2 was a mechanism where the rhythm of the music itself is generated by capturing the dancers' movements, a structure where a flow of sound that does not initially sound like a rhythm is gradually converted into one. However, due to the needs of the choreography, it ended up settling into a single rhythm. This means that between "rhythm being generated" and "rhythm for dancing," the center of gravity shifted slightly toward the latter this time.
Part 3 ── The Night Parade of One Hundred Demons and the Inhabitants of Another World
Part 3 was the final section we assembled for this performance, and as I was developing the new music, I structured it around the sounds of the tam-tam recorded on my piano album. It was also a choice to "trust the materials I had on hand" within a short preparation period.
Here, I will also open up a little about the system beneath the stage. Set 3 was originally designed as a recursive learning loop: having an AI learn Rosie's dance, a ghost dancing the new dance born from that, the dancer then receiving those movements as new choreography and embodying them, and the AI learning that once again── that was the design. In this interim report, we have established a part of that loop on stage by having the AI learn what the dancers performed over four days and incorporate that into a new dance. The theme of "inhabitants of another world" is embedded not just as a concept, but as a way of constructing the system itself.
The theme of the "Night Parade of One Hundred Demons" (Hyakki Yagyō) was born from Rosie being deeply influenced by the Hyakki Yagyō scrolls at the Tokyo National Museum. What I realized while talking with Rosie was that the grotesque figures depicted in the scrolls are by no means beings that negate humanity, but are depicted as inhabitants of another world who show us in a different form what humans have overlooked.
We chose an attitude of neither pessimism nor denial, even if some grotesque thing were to be born from our world, which the AI continues to observe. This might look like "suspending judgment," but I personally believe that affirming the grotesque is one of the concrete stances for living with technology from here on out.
Heading toward the Birmingham REP, this third part will likely be the section that undergoes the most significant transformation. Rosie's choreography will be further refined, and I am still in a period where I don't yet know how my music will respond to that.
Sound as a way of re-measuring distance
Sound as a way of re-measuring distance
When working on "Distant Echo," I wrote that I wanted people to experience "spatial audio as something where you choose the distance yourself." It was designed so that the audience could re-perceive time relatively while choosing their own distance from the speakers and performers.
I think what we aimed for in DEEP FAKE is fundamentally the same. Within our bodies, which continue to be observed and mapped by AI, I wanted to properly establish on stage, at least once, a moment where we can still choose our own distance with our own feet. The fact that there are moments within a mapped body where we still feel pleasure or interest ── I believe that those very moments might become the foothold for coexisting with technology. It is still a tentative answer, but that is what I think right now.
4 days of constraints, and looking toward Birmingham
For those who have read this far, I would like to share my honest thoughts.
For this interim report, we only had 4 days to assemble everything on-site. Including errors on the video side, we had to settle for what could be implemented within the limited time, and ultimately, we decided to operate by separating "sounds generated completely in real-time" and "pre-prepared audio" to a certain extent. To be honest, between the ideal of "everything being born within the space" that I aimed for at the design stage and the stable operation on the day, I shifted the center of gravity slightly toward the latter.
To put it frankly, I am left with a sense of frustration.
However, the fact that I can carry that frustration forward to the next stage also means that the work is not yet "finished." At Birmingham REP, I want to carefully rebuild the presentation of the ghost and system in Set 1, the rhythm generation in Set 2, the recursive learning loop in Set 3, and the synchronization with the video. I want to make it my homework for the next six months to ensure that the technology is not just running in the background, but is delivered properly to the audience's bodies.
Toward the next development
What's next
"DEEP FAKE: Ghost in the Machine" is scheduled to premiere at Birmingham REP in the UK in October 2026. After that, we will gradually grow the work through public performances in theaters in Japan, the UK, and Germany.
The content shown on 5/22 is not yet the final version. To move the answer forward, even a little, regarding whether the three times will truly become a "dialogue" or a "symbiosis," I intend to walk a little further with Rosie, the dancers, and everyone who is investing their time in this work.
To everyone who came to the venue and everyone who watched, thank you very much.
NAKAMURA Hiroyuki
https://nakamurahiroyuki.info/
all photo by Takahiro Kosaka
https://www.instagram.com/solinity?igsh=a25lYjNtam82YWVt
Credits
Co-production: Beyond Boundary Music (BBM) × K2CO Ltd (UK)
Choreography: Rosie Kay
Composition, sound, AI visuals: NAKAMURA Hiroyuki
Dancers: Ken Nakagawa, Yuko Shimoji, Saaya Takaoka
MC: Mayumi Taguchi (BBM)
Interpreter: Shibimi Matsuoka
Hosted by: Showa University of Music, Composition Department
Supported by: Arts Council England / The Great Britain Sasakawa Foundation
Photos: Takahiro Kosaka (@solinity)
#Work #InternationalExchange #Composer #Dancer #ShowaUniversityofMusic #OverseasActivities #AIandCreation #MediaArt #LivePiano #SpatialAudio #DEEPFAKE #GhostintheMachine #Somax2 #RAVE #composition #dance #AIart #spatialaudio