Say your neighbor's daughter, who's Deaf, stops by to return a casserole dish. Normally you'd both reach for a notepad, or she'd type out a sentence letter by letter while you wait. This time, she just signs toward your phone's camera — and the words appear on screen as she goes, no typing at all.
That's the idea behind Pixel sign-to-text American Sign Language translation, a new feature Google just built into its Pixel 11 phones.
What it actually is
Sign-to-text lives inside Gboard, the keyboard app already installed on most Android phones. It watches someone sign in American Sign Language (ASL) — the visual language many Deaf and hard-of-hearing people in the US use, built on hand shapes, facial expressions, and body movement rather than spoken words — through the phone's front camera, then turns it into written English sentences in real time.
The one step worth knowing first
If you or someone you know has a Pixel 11, here's the whole setup: open Gboard, add "Sign-to-text" to the keyboard toolbar, and grant Gboard permission to use the camera when asked. After that, tapping the Sign-to-text icon in any texting app, search bar, or even a prompt box for Google's Gemini assistant turns the camera on and starts translating as soon as signing begins.
Where you'd actually use it
For typing into an app — texting a family member, searching the web, or asking Gemini a question — Gboard's sign-to-text does the work. For a live, face-to-face moment, like ordering coffee or chatting with a neighbor, Google pairs it with Live Transcribe, the same tool already used for real-time captioning of spoken conversations.
What it can't do yet
I'll be honest about the limits here, since the support page is upfront about them too: this only works on a Pixel 11 right now, only translates from ASL into English, and needs the camera to get a clear view of your face, hands, and upper body to work well. If the lighting is bad or the phone is propped at an odd angle, it can lose track, so Google includes a short onboarding video showing how to position the camera before you start.
A quiet privacy detail worth knowing
Google says the camera doesn't send or store actual video — just a 2D map of hand, face, and body points, processed to generate text, then discarded.
If you're curious about other ways your keyboard is quietly getting smarter, Gboard can also clean up rambling voice typing into readable text, and if camera-based shortcuts interest you, Circle to Search can now identify several items in one photo at once.
Technology like this doesn't fix everything, but it's nice to see a phone meet people where their language already is, instead of asking them to type it out first.