Google Unveils SL2T AI for Real-Time Sign Language Translation

Sign-language users have long faced barriers when typing on mobile devices or communicating with people who do not understand signing, while speech-recognition systems offer limited help to deaf and hard-of-hearing communities. Google DeepMind’s SL2T model is designed to interpret signs captured by a smartphone camera and convert them into text, potentially expanding accessibility if it can deliver reliable performance across signing styles and real-world conditions.

Google DeepMind has now introduced SL2T with real-time text output from a phone’s camera feed. The initial rollout integrates the model into Gboard and the Live Transcribe app on the Pixel 11, targeting two core uses: composing messages through sign language and supporting in-person conversations. The launch brings sign-based input directly into Google’s mobile software, reducing reliance on spoken language or conventional keyboard entry.