Google is bringing sign-language-to-text translation into an everyday device. The Pixel 11 feature observes the signer through the front camera and converts American Sign Language into written English in real time. This is not simple hand tracking: the system also considers facial expressions and body movement, both essential grammatical elements of signed languages.

How Google ASL-to-text translation works

The feature can be added to the Gboard toolbar. A person frames their upper body with the front camera, signs in ASL and sees a written translation appear on screen. The output can be used in any text field or inside Live Transcribe, creating a direct communication channel between signing and non-signing people.

Google describes the underlying sign-language-to-text model as multilingual, while the supported product experience is ASL to English. It handles both one-handed and two-handed signing and interprets sequences rather than treating every gesture as an isolated word.

  • Input from the Pixel 11 front camera
  • Real-time translation into English sentences
  • Integration with Gboard and Live Transcribe
  • Combined analysis of hands, face and posture

Why this assistive technology matters

The most important innovation is not only the model itself, but where Google placed it. Building ASL into the phone keyboard lets people use their preferred language to compose a message, complete a form or communicate face to face without opening a separate app.

Google says the feature was developed in partnership with the Deaf community. That matters because a signed language is not a gestural version of a spoken one; it requires representative data, evaluation with native signers and interface choices shaped with the people who will use it.

Limits: there is no universal sign language

The feature does not automatically translate every signed language. ASL, Italian Sign Language and British Sign Language are different languages with their own vocabulary and grammar. Future expansion will require local data, community-led evaluation and support for regional variation.

Automated translation can also make mistakes. In medical, legal or emergency settings it is not a replacement for a qualified interpreter. Its immediate value is a new tool for everyday communication, provided its coverage and reliability are presented clearly.