· Built for everyday conversations

Sign, speak, read —
together.

Together translates ASL and Egyptian ArSL to text and speech — and back again — live, right in your browser. No install needed.

No install Private by default Bilingual UI Browser-native AI

( 01 ) — Our Products

Five bridges.
One platform.

Every direction of sign language communication — covered, in real time, with on-device AI that never leaves your browser.

( 02 ) — By the numbers

Accurate in
both languages.

250
ASL sign classes recognized
20
ArSL sign classes
<50ms
Avg. inference latency
100%
On-device privacy

( 03 ) — How It Works

From hands to meaning
in four steps.

1
Open browser

No download. No account needed for a demo. Just open Together in any modern browser.

2
Allow camera

Grant camera access. MediaPipe Holistic maps your hands, face, and pose — locally, never uploaded.

3
Start signing

The on-device AI classifies each sign in under 50 ms. Predictions stream as text and optional speech.

4
Connect & share

Join a live meeting session. Real-time captions appear for all participants across the room.

( 04 ) — Community

Voices from
our community.

★★★★★

"Together changed how I communicate at work. I sign, my colleagues read. No interpreter needed for quick questions."

SR
Sarah R.
Software engineer, Deaf/HoH
★★★★★

"As a teacher of the deaf, having ArSL recognized in the browser is incredible. My students use it every class now."

MK
Mohammed K.
Special education teacher, Cairo
★★★★★

"The VoiceBridge feature is stunning. I sign to my phone and it speaks for me at the pharmacy. Pure freedom."

AL
Alex L.
Accessibility advocate

( 05 ) — FAQ

Common questions.

Does Together require an internet connection?

The AI models run entirely in your browser using TFLite (English ASL) and PyTorch (Arabic ArSL). Once loaded, sign recognition works fully offline. Live meeting sessions require connectivity for WebRTC signaling only.

Is my camera feed sent to your servers?

Never. All video processing happens on your device using MediaPipe Holistic running in WebAssembly. Only anonymized text predictions can be shared in live sessions — and only when you choose to.

Which sign languages are supported?

Currently American Sign Language (ASL) with 250 sign classes and Egyptian Arabic Sign Language (ArSL) with 20 classes. We are actively training models for Gulf ArSL and British Sign Language (BSL).

How accurate is the recognition?

Under controlled lighting with good camera framing, ASL accuracy exceeds 92% top-1 across the 250-class vocabulary. ArSL reaches 89% on the 20-class set. Accuracy improves with a neutral background and sufficient light.

Can I use Together in a video call?

Yes. The Live Meeting module uses WebRTC via Socket.IO. Share a session link and captions appear for all participants live. Supports up to 8 participants per session.

How do I book a demo for my organization?

Visit the Book a Demo page or email us at hello@together.app. We offer 30-minute live walkthroughs for accessibility teams, educational institutions, and enterprise clients.

Ready to bridge the gap?

Start translating in seconds — no install, no signup required for the demo. Full features unlock with a free account.