Introducing Silent Speech interface for your existing devices! From your phone or computer, it lets you communicate with AI at the speed of voice, using only your mouth movements, no sound. Demo 0:13 and 1:09

While we were working on our AI operating system, we realized Voice is becoming one of the main ways we interact with AI and Agentic systems But the issue is, you can’t use voice everywhere, all the time. Not when you're in a loud environment, not when you want to talk privately, and obviously not when speaking would disturb the people around you, like in an office. Silent Speech solves that. It gives us the speed and naturalness of voice without requiring sound.

Mouth movements is very personalized, so we train a small custom model for every user. It takes just a few hundreds of sentences that you record, to get to a really high quality of silent-speech-to-text. You can join the waitlist and star training yours today - either through our guided online recorder, or through our Mac app, which is a regular voice-dictation app, but while dictate with your voice, it also records your mouth movements to train your Silent Speech model in the background.

Join the waitlist and start training your custom model today! Go to

I made something similar 2 years ago (macOS app wrapping a VSR model), from the users we had, we realized the novelty wore off quick and people just ended up whispering and stt was robust enough to capture whispers. People also felt uncomfortable needing to give webcam access. Curious to see how user feedback influences what you are building.

that is useful, thanks Chris!

bro this is actually insane 😭

It is!

Wait u r telling me I dont need the talk sock??

you, specifically, need to!

Let’s goooo, looking forward to trying it!

Very cool

Thanks Hubert!

Very cool!

Thanks Pablo! wanna possibly add it to Aqua?

Adam laying the groundwork for next generation of computer interaction!!

🫶! if you ever thinking about silent speech for pocket somehow, lmk

Cool primitive, but this only works if the webcam is on and pointed at your face constantly. That's a bigger ask than the audio problem it's solving for a lot of people.

It is a ask, true - but also the value is pretty good I'd say. using it on the go, in a bar, is a cool experience

1) What (!)

yeah pretty crazy!

😎

Personal pain point of mine and of a lot of other colleagues of mine! Can't wait to use it! I do wonder if the application learns specific movements of the mouth that the user has in order to improve the ability to transcribe.

yeah it needs you to record a few hundreds of sentences before !

already contributed 10 😅 Try to gamify it Add rewards and leaderboard

I can finally prompt in bed

without the wife killing you!

Insane Idea even to think about.

🫶

incredible

thanks Jared!

Does it only work on live input or can it read lips on vids too?

technically can read videos too

Politicians are screwed!

🔥

🫶

Very nice

🫶🫶

Very cool!

thanks Graham!

LFG, this is super cool

🤘

Windows n android users deserves love also

hahaha going to be on Android too, not so sure about windows! :p

This is what HAL used! People will die because of this!

😅😅😅

This is true accessibility! 🖤

true :)

Ooooh nice bro

thanks bro!

Super impressive Adam! Can't wait to test it

thanks François! you can start training your today :D

Brilliant!

thanks sanket!

though of this, nice to see this work!, what do you think is the max accuracy u can get to in the next few months?

based on the past few weeks, I believe i can get it to ~4% for most people who brings 1000+ sentences

oh wow, i looked up SOTA once and it was 80% something like that

baaaanger

thaaaankyouu

ngl, kinda dystopian

This is amazing!!!

yeah it is pretty crazy! when I woke up one morning and saw the WER goes to 8...7...6... I was in disbelief !

congrats!

thanks Richelle!

Wow 🤯 this is actually really useful!!!

yeah it is pretty dope this is written with my mouth movement only

INSANE

JAY!

That awesome lol