Google's Pixel 11 Can Turn Sign Language Into Text

Google launched the Pixel 11 family on August 12, 2026, and with it debuted a new DeepMind model called SL2T that transcribes sign language into text in real time. The model is built directly into Gboard, Google's default Android keyboard, and Live Transcribe, the company's existing real-time captioning tool for spoken audio. Deaf and hard of hearing users can now sign to their phone in situations where they would normally type, and the phone converts those signs into written text on the fly (Engadget).

At launch, SL2T supports only American Sign Language (ASL) transcribed into English. That is a narrower scope than the training data might suggest. DeepMind trained the model on more than 100,000 hours of multilingual sign language video, with roughly a quarter of that corpus in ASL. The simultaneous multilingual training was deliberate: by learning multiple sign languages at once, the model picked up shared structural patterns between them, a strategy that could speed up expansion to additional language pairs down the road.

The architecture has a deliberate privacy design. SL2T does not interpret raw video. Instead, a separate model running on the phone itself converts camera footage into a wireframe — a skeleton-like set of geometric coordinates representing the user's hand, arm, and body movements. Only that abstract skeletal data is sent to Google's servers for transcription. The server never receives identifiable video of the user's face or body. The wireframe approach also cuts bandwidth compared with streaming full video frames.

On the transcription side, SL2T skips a step that has been common in sign language recognition research. Traditional systems typically produce intermediate labels called glosses — essentially word-level tags for individual signs — and then translate those glosses into the target language. SL2T translates the wireframe data directly into text, eliminating the gloss layer entirely. This mirrors a broader trend in machine translation and speech recognition, where intermediate steps have been progressively removed in favor of models that go straight from input to output.

DeepMind announced the feature in a blog post titled "Putting sign language AI into users' hands" (DeepMind blog). The post coincided with the Pixel 11 hardware launch.

The scale of the potential audience is large. Globally, more than 70 million deaf and hard of hearing people communicate in roughly 200 sign languages. ASL-to-English covers one pair out of that 200. The multilingual training strategy suggests Google is laying groundwork for broader coverage, though the company has not announced a timeline for additional languages beyond what shipped at launch.

The decision to place SL2T inside Gboard and Live Transcribe is a deliberate distribution choice. Gboard is where Android users already type, and Live Transcribe is where they already read captions for spoken audio. Embedding sign language transcription into both surfaces means the feature shows up wherever a user would normally type or read captions, rather than being confined to a standalone accessibility app. That integration matters for adoption: accessibility features tucked away in dedicated apps tend to reach only the people who already know to look for them. Surfacing the capability in primary input and transcription tools gives it a substantially larger effective distribution.

The broader context here is the gradual convergence of on-device machine learning, cloud-based inference, and accessibility tooling. We have watched this pattern develop across the smartphone era: a capability that once required specialized hardware or dedicated software slowly migrates into the default operating system layer. Live Transcribe itself followed that arc when it launched for speech. SL2T applies the same logic to sign language.

Worth flagging is the architectural decision to split the pipeline across device and server. The on-device wireframe extraction handles the privacy-sensitive work of processing identifiable body imagery, while the heavier transcription model runs on Google's servers. This is a pragmatic trade-off. Full on-device processing would likely impose latency and heat constraints that could degrade the real-time typing experience Google is aiming for, while full server-side processing would mean transmitting raw video of the user. The wireframe intermediary is the compromise that keeps both privacy and performance in an acceptable range.

For developers and accessibility engineers, the gloss-free, end-to-end design is the detail most worth tracking. If SL2T's direct wireframe-to-text approach holds up in real-world use, it could influence how other teams build sign language recognition systems, particularly in settings where building a gloss dictionary for every language pair is impractical. The multilingual pretraining compounds that effect: adding a new target language should require less labeled data than training from scratch.

The Pixel 11 launch puts this into users' hands on day one. Whether it expands beyond a single language pair, and how quickly, will determine whether SL2T becomes a genuinely useful accessibility tool for the global deaf community or stays a capable but narrow demonstration of what the architecture can do.