Plain AI Daily

Pixel 11 Sign-to-Text Explained: Sign in ASL, Get English Text Anywhere You Type

By 8 min read

Google's Pixel 11 phones can now turn American Sign Language into written English in real time, using the front camera and the Gboard keyboard. It works in any app you can type in, it needs no subscription, and Google says your video never leaves the phone.

Google has shipped something genuinely new on the Pixel 11: a keyboard button that turns American Sign Language into written English while you sign, in any app you can type in. Google published the how-to on August 21, 2026, nine days after announcing the feature alongside the phones.

If you sign, this is the first time a mainstream phone lets you use your own language to write a text message, run a search, or ask Gemini a question. If you do not sign, the honest answer is that nothing on your phone changed today. Here is who gets it and what it actually does.

Key Takeaways

  • It is Pixel 11 only. Google's help article lists the prerequisites plainly: "Have a Google Pixel 11 device," Android 8 or later, the latest Gboard, and a keyboard language of en-US or en-CA.
  • There is no subscription. No plan is named in the prerequisites, in the how-to post, or in the launch footnote attached to this feature.
  • American Sign Language to English, and nothing else yet. Google DeepMind says "additional languages will follow" but names none.
  • It works anywhere you can type. Sign-to-text is a Gboard button, so it covers messages, notes, web searches and Gemini prompts. Live Transcribe also accepts signed replies.
  • Your video does not leave the phone. Google says the phone sends "only a 2D representation of key points of the movements, not your video," and that "your inputs are never saved, shared, or used to train models."
  • One hand is enough. Google specifically supports one-handed signing, for the obvious reason that your other hand is holding the phone.
  • Google publishes no country list. The launch footnote says only "available in select countries and languages."

Who Gets Pixel 11 Sign-to-Text

You need a Pixel 11. That is the gate that decides this for almost everybody, and Google states it as a hard prerequisite rather than a recommendation. The full list from Google's help article is short enough to check in a minute:

RequirementWhat Google says
Phone"Have a Google Pixel 11 device"
Android version"Be using Android 8 or later"
App"Install the latest version of the Gboard app"
Keyboard language"set keyboard language to en-US or en-CA"
PermissionGboard must have camera access, or the feature will not run
PlanNone named anywhere

The Pixel 11 lineup is Pixel 11, Pixel 11 Pro and Pixel 11 Pro XL. Google DeepMind says "more devices are coming soon," which is the only thing published about older Pixels, other Android phones or iPhones. No device is named and no date is given, so treat any specific claim about your phone getting it as unverified until Google says so.

Is It Free?

Yes, as far as anything Google has published goes. There is no Google AI Pro or Ultra requirement in the prerequisites, and the how-to post mentions no plan.

The strongest evidence is a detail in Google's own small print. The Pixel 11 features post of August 12, 2026 hangs a numbered footnote off each AI feature. The footnotes for Gemini Intelligence and for Rambler, the new voice input feature, both warn that "some features may require subscription for higher usage" and restrict use to people 18 and over. The footnote attached to sign-to-text says something different and shorter: "Results may vary. Check responses for accuracy. Available in select countries and languages." No plan, no age limit.

That is a meaningful contrast inside one document, not an absence of information. It still is not a positive statement that the feature is free forever, and Google has not published one.

What It Actually Does

Sign-to-text replaces typing, not conversation. You tap a button on the Gboard toolbar, the front camera opens with a preview window, and as you sign, English words stream into the text field you were about to type in.

Google's description of the mechanics is specific. The model reads "hand shapes, facial expressions, and body movements" rather than just hands, which matters because sign languages carry grammar in the face and body as well as the hands. Landmarks appear over your face, hands and upper body in the camera preview so you can see that you are framed correctly. Moving your hands out of the camera's view tells it you have finished.

You can mix signing with typing. Google says you can tap into the standard keyboard to edit a word by hand, pick alternatives from the prediction strip, or run the result through Gboard's writing tools to punctuate and polish it, without losing your place.

Where You Can Use It

Anywhere Gboard appears. Google's wording is "any app that accepts Gboard input," and the examples it gives are writing messages, taking notes, searching the web and prompting Gemini.

There is a second surface. Google says sign-to-text also works in Live Transcribe, the accessibility app that captions speech around you: instead of typing your side of a face-to-face exchange, you sign it. Google's own division of labour is that Gboard sign-to-text is for writing, and Live Transcribe is for "spontaneous face-to-face chats like ordering a coffee or talking with colleagues."

What Google Does With the Camera Feed

The camera runs, but Google says the video never leaves your phone. The phone tracks a set of points on your face, hands and upper body, and it is those coordinates, not the picture, that go to Google's servers for translation.

Google's exact wording in the help article is worth quoting because privacy claims about always-on cameras usually are not this specific:

  • "It sends only a 2D representation of key points of the movements, not your video, to the server and translates them into written English."
  • "These coordinates are processed using secure servers to perform real-time translation. Your inputs are never saved, shared, or used to train models."
  • "The pose coordinates, your videos, and translations aren't stored."

Google DeepMind's technical post describes the same design from the other end: an on-device model tracks the pose landmarks "allowing the original video to be discarded immediately," and only the geometric coordinates are sent for translation.

Two things this does not mean. It is not offline: the translation happens on Google's servers, so it needs a connection. And it is not a promise about other apps, only about sign-to-text.

Why This Is a Bigger Deal Than a Keyboard Feature

Because sign languages are not English on the hands. Google DeepMind's post makes the point directly: they are "independent, natural languages with their own distinct grammars and lexicons," which is why translation is a genuinely harder problem than speech-to-text and why earlier attempts, like sign language gloves, did not work.

The model behind this, called SL2T, was announced on August 12, 2026. Google says it was trained on more than 100,000 hours of data across more than 50 sign languages, with roughly a quarter of that in ASL, and that it scores 70 BLEURT zero-shot on FLEURS-ASL, a benchmark for ASL-to-English translation quality. Google calls it the first time sign language AI has reached consumer products.

Two practical details from the same post say more about whether it will work in real life than the benchmark does. Google says it worked specifically on "ensuring fairness for the 10% of signers who are left-handed," and on one-handed signing, "which is used while holding a smartphone in the other hand." Those are the details you only chase if real signers tested it.

Google says the feature was "developed in partnership with the Deaf community." It gives no further detail about who, which is worth knowing when weighing that phrase.

What Google Has Not Said

Four gaps, so you are not left guessing:

QuestionStatus
Which countries?Unpublished. The launch footnote says only "select countries and languages." The en-US or en-CA keyboard requirement is a keyboard setting, not a country gate.
Which other devices, and when?"More devices are coming soon." No device named, no date.
Which sign languages are next?"Additional languages will follow." None named, no date.
Is there an age limit?None published. The footnote for this feature omits the "18 and over" wording that Google attaches to its other Pixel 11 AI features.

Should You Care?

Get a Pixel 11 for this if you or someone you live with signs in ASL and currently types English on a phone all day. This is the first time the swap is available at all, it costs nothing on top of the phone, and it covers every app rather than one Google-shaped corner of the device.

Wait if you sign a language other than ASL, or you are on any other phone. Nothing is shipping for you yet, and Google has published neither a device list nor a date.

Ignore it if you do not sign. It is a keyboard option you will never turn on, and it does not change anything else about your phone.

The bottom line: Google put real sign language translation into a shipping phone, made it free, made it work in every app, and published a clear privacy boundary around the camera. The catch is the size of the door: one phone series, one sign language, one written language, and no published plan for widening any of the three.

Sources: Google's how-to post Here's how to use sign-to-text translation on Pixel 11 (August 21, 2026), Google's help article Use sign-to-text to translate American Sign Language (ASL) into English text in real time, Google DeepMind's Putting sign language AI into users' hands (August 12, 2026), and the Pixel 11 features post (August 12, 2026) including its numbered footnotes. All read on August 21, 2026. The help article's computer, Android and iPhone variants were each checked and are identical.

Keep up without the jargon

Frequently Asked Questions

Which phones get sign-to-text?

Only Google Pixel 11 phones. Google's help article lists the prerequisites as a Pixel 11 device, Android 8 or later, the latest Gboard, and a keyboard language of en-US or en-CA. Google says more devices are coming, but names none and gives no date.

Does Pixel 11 sign-to-text cost anything?

No subscription appears anywhere in it. Google's prerequisites are a Pixel 11, an Android version, the Gboard app, a keyboard language and camera permission, and that is the whole list. The launch footnote for this feature names no plan and no age limit.

Which sign languages does it understand?

American Sign Language only, translated into written English. Google says additional languages will follow but has not named one or given a date. Your Gboard keyboard language has to be set to en-US or en-CA for the feature to be available.

Where can I use sign-to-text on a Pixel 11?

Any app that accepts Gboard input. Google names writing messages, taking notes, searching the web and prompting Gemini. It is a keyboard button, so anywhere a keyboard appears, sign-to-text can appear instead of typing.

Does Google keep a video of me signing?

Google says no. The phone tracks points on your face, hands and upper body and sends only a 2D set of coordinates to the server, not the video. Google's help article states your inputs are never saved, shared, or used to train models.

How do I turn sign-to-text on?

Open the Gboard toolbar and tap Sign-to-text. If it is not there, tap Edit, then touch and hold the Sign-to-text icon and drag it onto the toolbar. You have to grant Gboard camera permission before it will work at all.

Get the plain-English AI brief

One email. What changed in AI and what it means for you.