Human-computer interaction, from paper to product.
Emerging science for how people see, hear, and speak — turned into working technology.
Two ways to start.
Unvoiced is a human-computer interaction studio in Sydney and Kyoto. Work reaches us in one of two ways.
One is a daydream. A thing we want to use that nobody has built yet. We sketch. We prototype. The science catches up to the wish. Proudly a solution in search of a problem. Imagination first, customer later.
The other is a real issue with a real deadline. A team arrives with a use case and a hunch that the science is almost there. We find what’s ready, and build what’s missing between the journal and the job.
Either way, we go the whole distance — paper, math, signal, embedding, enclosure, app. One team, no handoff. We finish what the paper started.
“Bold like an angel, careful like a devil.”— Kurosawa
And two ways to fund.
Sponsored research when the question is open and the horizon is long — you back the work, we keep the lab notebook public to you, IP shaped by agreement.
Commissioned work when the spec is sharp and the date is real — you own the deliverable, we ship it.
Most engagements live somewhere on the line between the two. We write the contract to fit the work, not the reverse.
Our main bet is Silent Speech Interfaces.
Voice is the fastest way people externalise thought. It’s also the loudest. In open offices, trains, libraries, bedrooms at night — most of the places people actually live — voice isn’t an option. And AI is making voice input more valuable by the month.
SSI closes the gap. Whisper, sub-audible articulation, vibrations from the body — acoustic signals quietly produced, that today’s microphones can’t reliably catch. We work multimodally, but keep the acoustic channel at the centre. You still hear yourself. You still shape breath into sound. The smallest audible voice, minimised but not erased — where the tool recedes into tacit use.
Whisper into your computer. Every app hears you.
A whisper microphone for your hand. Cup your palm and speak.
A tabletop stand that keeps its ear and eye on you.
Colour vision correction. Calibrated to your eyes, not an average pair.
A design engineer and a principal scientist, two time zones, one studio.
Unvoiced was founded by Sato Ozawa and Jun Rekimoto in 2024. Between Sydney and Kyoto, between product and paper — one team, no handoff.
Multiple-time founder working across hardware, software, and the seams between them. Writes the code, shapes the enclosure, argues with the signal. The studio’s generalist — and the one who ships.
Thirty years shaping human-computer interaction — multi-touch, augmented reality, and wearable input, before those words had settled into the language. Unvoiced is where the next chapter gets built.
Let’s talk.
A hearables brand chasing a new input modality. A medical device team with a hard signal-processing problem. A research group that needs the prototype their paper deserves. A neighbourhood shop wondering whether AI can actually help them. Different worlds, same craft. Sponsored research, joint development, or a scoped commission — the contract follows the work, not the reverse.