Etherea

Turns speech, music, and other live signals into projected visual environments intended to make shared spaces more responsive and socially engaging.

James Barnes

Poster

Key takeaways

Real-time visualization turns the room itself into a responsive interface.

Voice and music drive projection-mapped visuals across ceilings, instruments, and screens.

Local hardware enables self-contained operation, while remote inference can reduce latency during live shows.

Summary

Etherea was presented as an Imagination Engine for shared physical spaces. It pairs voice and music with real-time, interactive visuals that can be projected onto ceilings, instruments, and other surfaces.

The team said its first installation followed two years of research and development and was completed shortly before the talk in Nouriel Roubini's East Village penthouse. Across a series of hosted events, the team said it observed stronger attention and connection among participants; the talk did not present a controlled study of that effect.

The demonstrated machine was described as containing two RTX 5090 GPUs and capable of local operation, although the live system used Cerebras for fast language-model inference. Expansion to additional venues, a six-figure business trajectory, and mobile applications were presented as plans or forecasts rather than completed outcomes.

Why it matters

The room becomes part of the interface

Etherea explores an interaction model in which a gathering's shared environment responds to conversation and music, making generated media collective and spatial.

Limits & caveats

The reported increase in attention, presence, and connection is a team observation from installations and events, not a controlled behavioral result.
The first-installation timing, event count, and customer details are presentation-time claims and were not independently verified in the packet.
The six-figure trajectory, venue expansion, and application-store experiments are forecasts or roadmap items rather than completed milestones.
Although the machine was described as locally self-sufficient, the presentation explicitly said the live configuration still used an external inference service for speed.

How it works

Capture live speech and music

Speech, conversation, and music provide the live signals that drive the installation.

Select the inference path

The machine can run locally on two networked RTX 5090 GPUs; the live event used Cerebras for low-latency language-model inference.

Generate responsive visuals

The Imagination Engine turns the live input into responsive visualizations as it arrives.

Project across the room

Projection mapping places the visuals on ceilings, the piano, screens, and other physical surfaces.

Transcript

Imagine the living room of the future.

It entertains your guests, it dreams with your kids, it visualizes your music, it matches your energy, it adapts to whatever is happening, and it brings you into presence with it. It helps you connect with the people that you love, and somehow makes everyone inside of it, feel, like they are home.

I'm James, I'm with Gonzo, who is, also helping, run AV with me today. We founded Etherea, which is the Imagination Engine. We've been building, this room, and to today you're in one, this living room, where every word from every founder will be visualized on the ceiling in real time, from a machine called the Imagination Engine that's just right there. And by pairing your voice with interactive visualizations that can live on any surface, Etherea brings living rooms to life through living, breathing intelligence. After two years of R&D, we were proud to complete our first installation a few weeks ago at the East Village penthouse of Nouriel Roubini. We sold Nouriel, a famous economist and well-known super host, on turning his house into a living lab.

And we mapped his very white three-story penthouse with light, controlled by the voices of him and his guests. The ceilings came alive with color and the piano, the, baby white grand piano, was projection mapped explosively to music. And then there are screens on two floors like this that visualize the conversations of New York's most interesting people, like The Moth champion storyteller describing, living under the northern lights in Alaska, or the man with the record for the longest time spent alone at sea, which is three years, making his case to lead humanity to Mars, and, the creator of one of New York's best, creative gatherings playing our, grand piano and wondering, "How do I get this into the club where I do my gatherings every week?"

At one of New York's most iconic places to gather for decades, our host told us that we'd completely transformed the place over the dozen events that we'd held at his place over the past few months. And we're now seeking to expand it to more living rooms, knowing how universal the concept of a living room is. A living room is in a hotel lobby where people gather and talk and tell stories and bide time, in a mansion in the Hamptons, in a one-bedroom in Chinatown. A living room, is a place where people live and gather and tell stories. And through real-time visualization, we have found that we can increase the amount of attention people pay to each other, building deeper presence, connection, and imagination.

And we believe that it's an important new medium with an unlimited potential of real-time visualization. Nouriel now wakes up to personalized visualizations of Bloomberg News, as an example. We're on track to build a six-figure business through installations and events, and we're focusing our growth efforts on buying new machines, and putting them in new living rooms in places like hotels, venues, and so on.

So apparently the, school has a bunch of local AI. I was wondering, does Etherea run in the cloud, or does it use any local AI?

Thanks to our friends at the Green Point Compute Co-op, we've now, So this machine here has two RTX 5090s, that are networked together, and this is called Tulip, and the first generation of the machines which we sold to, Nouriel was called Daisy. And so we have, made this machine which can, yeah, run just completely self-sufficiently. Right now it's using Cerebras for LLM inference because it's super fast, which really matters when you're putting on a show. But if you want to use it on a desert island should you have power, you could totally do that. Or in an auditorium, I guess, in an old school. Should you have power, you can do it.

And then, we have, we have, we're, experimenting with a few mobile apps which are running on RunPod now. So, I think we'll start to ship some stuff into the App Store and run some experiments. I think our business model is gonna be, really we're gonna try and sell these things into into the space as we described. I think, you know, having GPUs in people's houses is a good thing, for all of the reasons that people in the cohort are describing, and also just to be like self-sufficient and have an item of beauty. So I think we try to like use the best of what we have, and our goal is to allow anybody who, wants to buy our technology to use it completely with just the electricity in their house.