Understand what was signed
Camera motion becomes language-neutral frames, then recognition maps sequences to written meaning for that sign language — DGS, ASL, BSL, Egyptian Sign Language, and more.
NEOLANG
ecosystem
SignToTalk
Any sign language. Both directions.
Not a learning site — a platform you wire into products. Signed motion becomes text, and text becomes signing motion, with perception that stays on the device.
- On-device perception
- Motion ↔ text for any sign language
- Subscriber API keys for every client
The core
One engine. Two directions.
SignToTalk digests any sign language for software — not a study catalogue. Signed motion becomes text; text becomes signing motion your product can deliver.
Express meaning as signing
Written or spoken language maps back into articulator motion — for avatars, accessibility flows, and client experiences that need signed output.
How the platform works
From camera to shared language.
Clear layers. Stable contracts. Privacy at the edge. Each step can improve without breaking clients.
01
Perceive
On-device body, hands, fingers, head, and face as UniversalMotionFrame.
02
Digest
Engine maps motion ↔ text for any supported sign language.
03
Deliver
Text, speech, APIs, and embeds into client platforms.
04
Express
Mapped motion drives a signing avatar in your product.
Available now
Build on what already ships.
Perception, contract, embed, and articulator avatar are live. Bidirectional motion ↔ language is the product core we are building toward — not a side feature.
Trust
Video stays with the person.
Tracking runs in the browser. This host serves the client and the contract — it does not become a camera cloud.
- No camera frames or video uploads for tracking
- No MediaPipe or RTMPose on the droplet
- No signing video stored here