---
title: "What Are Subvocal's Silent-Speech Glasses and Do They Work?"
url: "https://goremy.ai/blog/articles/subvocal-glasses-silent-speech-radar-499-preorder"
author: Luis Chavez-Mattos
editor: Luis Chavez-Mattos
editor_title: Director, Product
published: "2026-11-03T02:20:05.726Z"
updated: "2026-11-03T02:20:05.726Z"
topic: AI Industry & Strategy
description: "Subvocal's radar-based glasses claim to turn silent mouthing into text at 200+ wpm for $499 pre-order. Here is how they work and what remains unproven."
---

# What Are Subvocal's Silent-Speech Glasses and Do They Work?

> Subvocal's radar-based glasses claim to turn silent mouthing into text at 200+ wpm for $499 pre-order. Here is how they work and what remains unproven.

Edited by Luis Chavez-Mattos, Director, Product · Published 2026-11-03

## What are Subvocal's silent-speech glasses?

Subvocal's glasses are a prototype wearable from a small Y Combinator startup founded by two Georgia Tech students. They send radar signals toward the face and use AI to turn tiny lip and jaw movements into text, so you can "type" by silently mouthing words. The company claims speeds above 200 words per minute, compared with about 40 for typical typing. Pre-orders cost $499. The speed claims have not been independently tested.

## TL;DR

- Subvocal is a **Y Combinator startup founded by two Georgia Tech students**, and its glasses are still a prototype.
- The glasses use **radar to sense lip and jaw movement**, and an AI model converts that motion into text.
- The company claims **over 200 words per minute**, roughly five times the 40 wpm typical for typing.
- **Pre-orders are open at $499**, but a prototype with a pre-order is a bet, not a finished product.
- Apple reportedly paid **$1.6 billion earlier this year** for a company with similar technology, which signals serious interest in the category.
- **No independent speed tests** have surfaced yet, so the headline number is the company's own claim.

## How do the glasses work?

The core idea is that speaking, even silently, involves small physical movements. Your lips, jaw, and surrounding tissue move when you form words, even if you make no sound. Subvocal's glasses emit radar signals toward the face and read how those signals change as the muscles and skin move.

An AI model then maps those movement patterns to words. The company's pitch is that this works without a microphone picking up your voice, so the input is silent and private.

Radar is the notable choice here. Cameras need line of sight and decent lighting, and microphones need audible sound. Radar can sense small motion without either. What the source material does not tell us is the radar frequency, the sensor count, the model architecture, or the training data. Those details matter for judging accuracy, and they are not public in the material we have.

## What can you do with them?

The demo scenarios center on discreet interaction with an AI assistant and with other people:

- **Silent AI commands.** In the demonstration, a user silently asks an AI assistant to summarize Slack messages while sitting next to coworkers, and nobody notices.
- **Silent typing.** The glasses are meant to turn mouthed words into text at speeds well above normal keyboard typing.
- **Private conversation.** The company says two people wearing the glasses could talk across a room without anyone else realizing.

The last claim is the most ambitious. It implies a transmission path between two wearers, not just local text entry. The sources do not explain how that would work, so treat it as a stated goal.

## How much do Subvocal's glasses cost?

Subvocal is taking pre-orders at **$499**. The product is still described as a prototype. The sources do not give a shipping date, a refund policy, battery life, or what the price includes, such as a subscription for the AI features. Check those terms before paying.

| Item | Detail |
|---|---|
| Pre-order price | $499 |
| Status | Prototype |
| Claimed speed | 200+ wpm |
| Typical typing speed cited | about 40 wpm |
| Shipping date | Not stated in sources |

## Do the speed claims hold up?

Not yet verifiable. The 200 wpm figure comes from the company, and as of the source material nobody has independently tested it. That gap matters for several reasons.

- **Speed is not accuracy.** A system can output text quickly and still get many words wrong. A useful number would pair speed with error rate.
- **Test conditions matter.** Scripted phrases, limited vocabularies, and trained users all produce better numbers than open-ended speech from a new user.
- **Demos are curated.** A smooth demo shows the best case, not average performance across face shapes, accents, and movement.
- **Mouthing is not thinking.** Calling this "telepathy" is a marketing frame. The system reads physical movement from your face, so you still have to form the words.

For context, 200 wpm is faster than most people can speak aloud in normal conversation. That does not make it impossible, but it is a high bar and deserves independent measurement.

## Why does Apple's $1.6 billion purchase matter?

Earlier this year, Apple reportedly paid $1.6 billion for a company that built very similar technology. The source material does not name the company or describe the deal terms, so we cannot compare the two technologies directly.

What the deal does suggest is that a large hardware maker sees silent or near-silent input as worth real money. It is a signal about market interest, not proof that Subvocal's product works. A big acquisition elsewhere in the category does not validate a specific startup's accuracy claims.

## Should you pre-order them?

It depends on your tolerance for risk. A $499 pre-order on a prototype means you are funding development. Consider these questions first:

1. Does the company publish accuracy numbers, not just speed?
2. Will independent reviewers get units before you commit?
3. What happens to your money if the product slips or never ships?
4. Where does the facial radar data go, and is processing done on the device or in the cloud?
5. Would you actually wear them in public, mouthing words at your desk?

The privacy question deserves attention. Any device that reads your facial movement and feeds an AI assistant handles intimate data, and the sources say nothing about how it is stored or processed.

## Frequently Asked Questions

### Who makes the Subvocal glasses?
Subvocal is a small Y Combinator startup founded by two Georgia Tech students. The glasses are a prototype, and the company is already taking pre-orders.

### Do the glasses read your thoughts?
No. They use radar to detect the small movements of your lips and jaw as you silently mouth words, then use AI to convert that motion into text. You still need to form the words physically.

### Is the 200 words per minute claim confirmed?
No. It is the company's own claim, and no independent testing of it had been reported at the time of the source material. Typical typing is cited at about 40 wpm.

### How much are the Subvocal glasses?
Pre-orders are $499. The sources do not give a shipping date or say whether any ongoing fees apply.