01
Motion Capture
The camera records gesture input at 60 fps, isolating hands from the background in real time.
Welcome to GESTURELINK
A Real-Time Hand Gesture-to-Text System for the Hearing and Speech Impaired
Live viewport
Watch signing become sentences
Skeletal tracking overlays every joint while text streams out the moment a sign resolves.
Live camera feed (simulated)21 keypoints · 12 ms inference
Transcription
Hello, my name is Aria. Nice to meet you!|
How it works
Three steps, milliseconds apart
02
AI Landmark Mapping
A neural network maps 21 hand keypoints per frame, tracking joint angles and motion trajectories.
03
Instant Translation
Sequences are decoded into natural language and delivered as text and synthesized speech.
Capabilities
Engineered for real conversation
Zero-latency inference
Sub-15 ms on-device prediction keeps conversation at human speed — no cloud round-trip.
Multi-dialect library
ASL, BSL, ISL and regional variants, with a growing community-contributed sign corpus.
Offline functionality
The full model runs locally, so translation works in classrooms, clinics and transit alike.
Two-way communication
Speech is transcribed back into captions and animated signing avatars for true dialogue.
Mission & impact
Independence should never wait for an interpreter
Over 70 million people sign as their first language, yet most everyday counters, clinics and classrooms cannot answer back. GESTURELINK puts a translator in every pocket — so a hospital visit, a job interview or a conversation with a neighbour happens on equal footing, without a third person in the middle.
- 70M+
- people signing worldwide
- 12 ms
- average translation latency
- 100%
- on-device, private by default