
Hey everyone 👋
We’re Aarush & Arnav - co-founders of The Subvocal Company
TL;DR: We’re building glasses that let you talk to your computer without talking out loud. You mouth the words silently, and almost invisibly, and Subvocal decodes it to text in real time.
Edit: Since launch, we’ve received roughly 750 paid preorders and collected about $150,000 in refundable deposits. The launch video reached more than 12 million views on X. We’ve also miniaturized the prototype into more comfortable, portable glasses and are getting ready to ship it out to our first customers in a few months.
A more crude version of this won the YC G1 product showcase a few weeks ago!
Why we’re building this
Voice is the fastest, most natural way to use a computer and as agents handle more work, our job moves upstream, to structured and detailed prompting. Expressing your thoughts is best done through unconstrained, open voice. Wispr Flow’s popularity has shown how much people love fast input.
But voice also fails in exactly the moments it needs to work. In an open office or a cafe or in public transit, it’s incredibly awkward for you and distracting for those around you to talk into your computer.
People have found incredibly creative and weird workarounds to this, including gooseneck whisper mics and even sound proof masks you can talk into.
We wanted the speed of speech without having to make a sound.
How it works
Subvocal uses RF sensing from a pair of glasses to pick up the tiny facial movements you make while silently articulating words.
We send low-power radio waves through the face and measure how they change as your muscles move. Our models then decode those signals directly into text.
You can articulate clearly, or do it subtly enough that someone sitting across from you barely notices.
Where we are today
We started working on this about four months ago after building the first version at a hackathon.
Since then, we’ve gone from a table full of electronics and antennas taped to our face to a semi-portable board and sleek glasses that can decode open-ended speech at 200+ WPM and a reasonably low error rate even at a small dataset scale.
The hardware currently looks a lot worse than the product we intend to ship. Our website shows what the final product looks like, and we’re getting there fast. In the next week, we’ll miniaturize our hardware enough to fit completely on the glasses.
The core thing we wanted to prove first was that the signal exists and that we can decode it. Over the next few months, we intend to collect 000s of hours of population-scale RF data to pretrain a model that works with a very short calibration on anyone.
What we want this to become
We think silent speech can become the next interface for interacting with computers, replacing voice and the keyboard.
Instead of choosing between typing and talking out loud, you should be able to just silently articulate what you are thinking, and have it appear on screen.
Emails, prompts, messages, code, notes, commands, all at the speed of speech without bothering anyone around you.
Our goal is to get this into normal-looking glasses you can wear all day.
We’re opening reservations
We’re starting with people who already use voice heavily and wish they could use it in places where speaking out loud doesn’t work. Reserving early gets you a founders price and earliest batch shipping.
You can reserve one of our first units for $199 here: https://subvocal.co/preorder for shipping in the first limited batch and get a special $200 early bird discount on the final price.
We’re initially opening up a limited number of reservations while we work toward the first production run.
If you use tools like Wispr Flow, Superwhisper, ChatGPT voice, or dictation constantly, we would especially love to hear from you.
Our asks
1. Reserve one if you want this. The most useful signal for us right now is whether people are willing to put real money down for it.
2. Tell us where you would use it. Office? Meetings? Coding? Walking around? Somewhere else entirely?
3. Send this to the most voice-addicted person you know. We want to find the people for whom this immediately feels indispensable.
We started building Subvocal four months ago, and we’re moving as fast as we can to get this into your hands.