About
NVIDIA Signs is a free web-based platform designed to make American Sign Language (ASL) accessible to anyone, anywhere. Powered by AI and computer vision, it uses a learner's webcam to detect hand and finger positions in real time, providing instant feedback as users practice new signs. The platform features a structured learning path — starting with ASL basics — guided by a 3D avatar that demonstrates correct form and pacing. Beyond learning, Signs invites community contribution: users can record themselves performing specific signs to help build the world's largest open ASL dataset. This crowdsourced dataset captures the natural variation in how humans sign, and will be made publicly available so third-party developers can build accessible tools for the Deaf community. The platform targets two key audiences: learners (especially hearing parents of Deaf children who want to communicate more effectively) and contributors (both beginner and experienced signers who want to give back). Signs was created as part of NVIDIA's AI for Good initiative, in collaboration with Hello Monday/DEPT® — the team behind Fingerspelling.xyz — and the American Society for Deaf Children. Currently focused on hand and finger positions, the platform is expanding its dataset and curriculum over time. It is entirely free to use, with no registration wall for learning or contributing. All submitted video recordings are subject to usage terms that allow NVIDIA to publish the data for research and application development.
Key Features
- Real-Time ASL Feedback: Uses your webcam and AI-powered computer vision to detect hand and finger positions and provide instant feedback as you practice signs.
- Structured Learning Levels: Guides learners through ASL basics via a leveled curriculum, complete with a 3D avatar demonstrating correct sign form.
- Community Sign Contribution: Users can record themselves performing specific signs to contribute to an open, publicly available ASL dataset hosted by NVIDIA.
- Open ASL Dataset Initiative: All contributed recordings are aggregated into the world's largest ASL dataset, made freely available for third-party developers building accessibility tools.
- Multi-Camera Support: Automatically detects multiple connected cameras and lets users choose the best one for accurate sign recording and recognition.
Use Cases
- Hearing parents learning basic ASL signs to communicate with their Deaf or hard-of-hearing children at home.
- Students and beginners exploring American Sign Language through structured, self-paced AI-guided lessons.
- Experienced signers contributing video recordings to help build a diverse, open ASL dataset for global research.
- Accessibility researchers and developers leveraging the publicly available ASL dataset to build sign-language recognition tools.
- Teachers and educators supplementing classroom ASL instruction with interactive, real-time AI practice sessions.
Pros
- Completely Free: No subscriptions or paywalls — both learning and contributing are fully free, making ASL education accessible to everyone.
- Real-Time AI Coaching: Instant webcam-based feedback removes the need for a human instructor and allows learners to practice at their own pace.
- Community-Driven Impact: Contributions go toward an open dataset that enables developers worldwide to build accessible technology for the Deaf community.
- Backed by NVIDIA & Accessibility Experts: The platform benefits from NVIDIA's AI research capabilities and is developed in partnership with the American Society for Deaf Children.
Cons
- Hand/Finger Focus Only: The current AI model primarily recognizes hand and finger positions and does not yet cover non-manual signals (facial expressions, mouth movements) that are essential to full ASL fluency.
- Requires a Webcam: Real-time learning and sign recording both depend on a functioning webcam, which may limit access for some users or mobile devices.
- Limited to American Sign Language: The platform currently supports only ASL and does not cover other sign languages used globally (e.g., BSL, Auslan, LSF).
Frequently Asked Questions
Is NVIDIA Signs free to use?
Yes, NVIDIA Signs is completely free. Both the ASL learning modules and the sign contribution features are available at no cost to any user.
How does the real-time AI feedback work?
The platform uses your webcam along with computer vision AI to analyze your hand and finger positions as you sign. It compares your movements to reference signs and provides instant feedback to help you improve.
Who is Signs designed for?
Signs serves two groups: learners — especially hearing parents wanting to communicate with Deaf children — and contributors, who can record their signs to help build the open ASL dataset regardless of their skill level.
What happens to the videos I record and submit?
Submitted recordings are used to build NVIDIA's open ASL dataset, which will be made publicly available to enable third-party developers to create accessible applications for the Deaf community. All submissions are subject to the platform's terms of use.
Who built and supports NVIDIA Signs?
The platform is led by NVIDIA as part of its AI for Good initiative, built by Hello Monday/DEPT® (creators of Fingerspelling.xyz), and developed in collaboration with the American Society for Deaf Children.
