Real-time ASL → voice

The deaf community can hear you. Now you can hear them back.

Every accessibility tool turns speech into text. None turns sign into speech. VocalEyes watches a signer through any camera and speaks for them — out loud, in the moment, with no interpreter on call.

Hello
Spoken aloud from sign
Scroll
The gap

There aren’t enough interpreters. There never will be.

0
People in the U.S. who sign ASL as their primary language
Against roughly ten thousand certified interpreters, that’s a 50-to-1 ratio — and it isn’t closing.
~0
Certified ASL interpreters to serve them
$00/hr
Cost of a scheduled interpreter — booked days in advance

Interpreters get booked for the planned moments — the lecture, the doctor’s appointment. VocalEyes covers the unplanned ones: the walk-up question, the emergency room at 2am, the conversation that wasn’t on a calendar.

The wedge

We own the direction no one else points a camera at.

A conversation has two directions. Toggle between them.

The crowded half
Speech → text

XRAI, live captioning, transcription apps. A solved, crowded space that helps a deaf person understand a hearing one. Their microphone faces the hearing speaker — there is no camera pointed at the signer, so reading sign language is architecturally out of reach.

The empty half — ours
Sign → voice

A deaf person signs; the hearing person hears a voice. The other half of the conversation — the half that lets the signer be heard, not just informed. No competitor has a camera on the signer. VocalEyes is built around one.

See it work

A camera, a sign, a voice — no setup.

Real-time recognition running today on a core vocabulary, expanding through data collected directly with the deaf community.

Demo video goes here

Record the 60-second sign-to-voice demo, then swap this block for the embed. (Instructions are in the page source.)
How it works

Built on the one asset competitors can’t copy.

01
See

Any standard camera tracks the signer’s hands and posture frame by frame — no gloves, no sensors, no special hardware.

02
Recognize

A sequence model reads the motion of a sign as it happens and maps it to meaning in real time.

03
Speak

The recognized sign is voiced aloud instantly, so the hearing person responds to the signer — not to a screen.

04
The moat

The model isn’t the defensible part — the data is. Vocabulary grows through signing data collected with deaf-community trust, a position that compounds and can’t be scraped.

The business

Compliance budget today. Hardware standard tomorrow.

Now — B2B SaaS
ADA-obligated institutions
Universities, hospitals, and government already carry a legal duty to provide access — and a budget to match. VocalEyes fills the on-demand gaps interpreters can’t, starting at $299/mo per location. First beachhead: the SJSU accessibility office.
Next — Licensing
“Dolby for accessibility”
As AR glasses ship with always-on cameras, VocalEyes becomes the sign-recognition layer licensed per unit to manufacturers — the standard that runs underneath, not a separate app to download.

Say hello — with your hands or your voice.

Aarnav Zinzuwadia
Founder · VocalEyes

Data Science student at San José State University and solo founder of VocalEyes, currently building through the SJSU SpartUp summer intensive. If you’re an accelerator, an accessibility leader, or someone who cares about this problem — reach out directly.

Copied