Details
- Google DeepMind says SL2T is a sign language-to-text model now powering new accessibility features for Deaf and hard of hearing users on Android.
- The first rollout supports American Sign Language to English on Pixel 11, letting people sign directly into Gboard and Live Transcribe instead of typing.
- DeepMind says SL2T translates simultaneous movements of the hands, body, and face into text, reflecting how sign languages work in practice.
- The system is optimized for real-world use cases such as one-handed signing while holding a phone.
- To reduce privacy exposure, body pose tracking happens on-device while Google’s servers handle translation into text.
- DeepMind says the model was built with Deaf Googlers and its AI Sign Language Advisory Committee, and it is meant to expand to more sign languages and applications.
- Google describes SL2T as state-of-the-art on academic benchmarks and says it is its first consumer deployment of this sign-to-text capability on phones.
Impact
This puts Google into a more visible race to turn sign-language recognition into a mainstream mobile input method, a space that has long been limited to research prototypes and smaller apps. By shipping inside Gboard and Live Transcribe on Pixel hardware, Google is using distribution and accessibility integration as its main advantage, while its on-device pose tracking approach addresses privacy concerns that can slow adoption. The move also signals that sign-language AI is shifting from lab demos toward practical phone use, where translation quality, latency, and inclusivity will matter most.
