Google has released a new AI tool that converts sign language into text in real time, making everyday communication between deaf and hearing people significantly easier. The technology, called Sign Language-to-Text (SL2T), is now available as a standard feature on Google Pixel 11 devices released in August 2026. Researchers at Rochester Institute of Technology's National Technical Institute for the Deaf collaborated with Google over four years to test, refine, and evaluate the system before its public launch.

How Does the AI Actually Read Sign Language?

Unlike speech-to-text systems that only listen to audio, SL2T analyzes multiple layers of visual information simultaneously. The system reads hand shapes, finger positions, and movements, but also captures facial expressions and body movements, which are essential to conveying meaning in sign language. Google trained the model on more than 100,000 hours of sign language video, with approximately 25,000 hours dedicated specifically to American Sign Language.

The technology works with just one free hand, meaning users can hold a phone in one hand while signing with the other. Once the AI converts the sign language into text, users can review and edit their message before sending it, just like they would with a voice-to-text message. This mirrors the workflow that hearing users already know, making the tool intuitive for deaf and hard-of-hearing individuals.

What Makes This Different From Previous Accessibility Tools?

For decades, deaf individuals had limited options for quick communication with hearing people. Gary Behm, director of the NTID Center on Access Technology at Rochester Institute of Technology, recalls growing up with a pen and paper as his primary communication tool. Mobile phones brought typed text conversations, but SL2T represents a fundamental shift: users can now communicate in their native language, American Sign Language, without translating to English first.

"Sign Language-to-Text has the potential to expand communication access and empower deaf individuals around the world. This really is a breakthrough," said Gary Behm, director of the NTID Center on Access Technology.

Gary Behm, Director of the NTID Center on Access Technology at Rochester Institute of Technology

The system currently supports more than 50 international sign languages, though American Sign Language presented unique challenges. Unlike English, American Sign Language does not follow English grammar structure and uses spatial relationships and body movement to convey concepts. Google's engineers had to train the AI to understand these linguistic differences, which is why American Sign Language received such intensive focus during development.

Steps to Using Sign Language-to-Text on Your Device

• Device Requirement: The feature is available on Google Pixel 11 devices released in August 2026 as a standard feature requiring no additional download.
• Camera Activation: Open the camera app and position yourself so the phone can see your hands, face, and upper body movements clearly.
• Text Review: After the AI converts your sign language to text, review the message and make any edits before sending, just like with voice-to-text features.
• One-Handed Signing: You only need one free hand to sign; the other can hold the phone or be used for other purposes.

Behm emphasizes that while SL2T works well for everyday conversations, it is not intended to replace qualified human interpreters in high-stakes situations such as medical appointments, legal proceedings, or educational settings where accuracy and nuance are critical.

Who Helped Build This Technology?

The four-year collaboration between Google and Rochester Institute of Technology was instrumental in creating a system that actually works for real users. Sam Sepah, an alumnus of RIT and NTID, now serves as the lead AI accessibility research program manager at Google. Sepah founded Google DeepMind's AI Sign Language Advisory Committee, which brings together deaf community leaders, researchers, and subject matter experts to guide the responsible development and deployment of the technology.

"Arriving at RIT as a deaf student driven to make a difference, I built a strong foundation in leadership. Through the incredible support I received there, I gained the skills that eventually led Google to select me to build their Sign Language AI Research team," said Sepah, who earned a bachelor's degree in multidisciplinary studies in 2006 and a master's degree in human resources in 2008.

Sam Sepah, Lead AI Accessibility Research Program Manager at Google

The advisory committee's role is crucial for ensuring that new AI-powered communication tools reflect the values, needs, and expectations of the deaf community. This approach represents a shift in how tech companies develop accessibility features, moving from top-down design to collaborative development with the communities they serve.

What Are the Real-World Implications?

For deaf individuals whose preferred mode of communication is American Sign Language, this technology eliminates a significant friction point in daily interactions. A deaf person can now communicate with a hearing person without needing to type, without needing an interpreter, and without the other person needing to know sign language. The technology is faster than typing and preserves the natural flow of conversation.

The project also demonstrates how AI research benefits from genuine collaboration with the communities it serves. By involving deaf researchers, educators, and community members throughout the development process, Google created a tool that actually addresses real communication barriers rather than solving a problem that deaf people didn't have. This model of inclusive AI development could influence how other tech companies approach accessibility features in the future.