Silent speech glasses from The Subvocal Company are meant to turn words you mouth without sound into text on your computer. The San Francisco startup’s preorder terms took effect on October 30, 2026, for US buyers. They list an Early Bird price of $499 and estimate first shipments in Q1 2027.
This piece covers what Subvocal says the glasses do, how they reportedly work, the preorder terms and what the privacy policy covers. It ends with questions an IT lead should ask before staff bring a pair to work. Facts are as of November 2, 2026, from Subvocal’s own pages unless we say otherwise.
The short version: Subvocal claims text input at more than 200 words per minute from silent mouthing, with no camera. Nobody has independently verified that speed. The product is still in development, delivery depends on FCC approval, and the current privacy policy covers only the website and preorders. How the glasses will handle your words is not yet documented.
What Subvocal says its silent speech glasses do
Subvocal’s homepage tagline is "Talk without talking." The company describes the product as a way to "input text at the speed of speech, silently and invisibly."
The pitch is about places where talking out loud is awkward. The Subvocal homepage names trains, quiet rooms and offices with coworkers nearby. It lists messages, notes and "a conversation with AI" as uses. One section shows a user sending instructions to a coding agent at a shared desk.
According to the homepage FAQ, Subvocal is "being built as an input method, not as a separate destination app." The company says you "should be able to" use it with ChatGPT, Claude, Cursor, Slack, email, documents and messaging. That is a stated goal. The company has not published a list of supported operating systems or apps.
The preorder page describes the product as "glasses being built for silent text input into your computer at speaking speed." It also says the images show "our current design direction; final specifications are still being developed."
How the silent speech glasses work, as far as we know
Subvocal’s own pages say little about the sensing method. The homepage FAQ confirms three points:
- No camera. “Subvocal does not use or require cameras for silent speech input.” There is no camera-based lip reading, and your webcam stays off.
- Deliberate mouthing. You mouth words on purpose, “as subtly as you want,” without audible speech. The company says someone close by may notice slight movement.
- Calibration. There “will be a short personal calibration process” at setup. The company hopes to make it lighter over time.
The sensor itself comes from press coverage. Digital Trends reports that, according to the company, the glasses "send radar signals through your face." Silently forming words changes how those signals reflect. An AI model then turns the changes into text, Digital Trends says. We did not find the word radar on Subvocal’s homepage, preorder page or legal pages as of November 2.
The company also says what the glasses do not do. Asked whether Subvocal can read thoughts, its FAQ answers no. It "only decodes words you intentionally articulate." Asked whether it is always listening, the company again says no. Input happens when you choose to mouth something.
Because the glasses do not listen for your voice, the company says background noise does not cause the same problems as voice dictation. No outside test has checked that yet.
Speed and accuracy: claims, not test results
Subvocal says it "decodes almost-invisible subtle, silent mouthing into text at 200+ WPM." The homepage repeats the figure as "200+ words per minute."
Treat that number as the company’s claim. Digital Trends notes that the speeds "haven’t been independently verified." Subvocal has not published a test method, a vocabulary size, a word error rate or the number of people tested.
On accuracy, the company’s FAQ says the current prototypes "already decode open-ended silent speech in real time." It adds that accuracy "continues to improve as we collect more data and train larger models." It also concedes that the system "can occasionally get something wrong" and relies on context and easy corrections.
Digital Trends raises the open questions we would ask too: how the glasses handle different accents, unfamiliar words and long stretches of silent dictation. Ask for results on your own vocabulary, including names and acronyms, before you plan around the speed claim.
Price, preorder terms and timing
Subvocal’s preorder terms, effective October 30, 2026, set out the money and the timeline for the silent speech glasses.
| Item | What the company states | Source page |
|---|---|---|
| Early Bird product price | $499 per pair (preorder page shows $699 struck through) | Preorder terms and preorder page |
| Deposit | $199 per pair, credited in full; $300 balance due later | Preorder terms |
| Refunds | Full deposit refund any time before final payment; refunds initiated within seven business days | Preorder terms |
| Extras | Taxes, shipping, optional prescription lenses and accessories cost extra | Preorder terms |
| Where | US delivery addresses only; buyer must be 18 or older | Preorder terms |
| Timing | Estimated Q1 2027, “not a guaranteed delivery date” | Preorder terms |
| Regulatory | “Delivery subject to FCC approval” | Preorder page |
The balance is not charged automatically. You get a separate checkout link and 30 days to pay. Return policy and warranty terms come before final payment.
The terms also say "final specifications, appearance, materials, fit, software compatibility, features, packaging, and accessories may change." Anyone buying for a team should read that line as the current status of the product.
What the privacy policy covers today
This is the key point for IT and privacy staff. Subvocal’s privacy policy, effective October 30, 2026, covers the website, preorders, referrals and related messages. It does not yet cover the glasses.
The policy says a future product or service that "processes voice, sensor, device, or other product data will receive a separate or updated privacy notice before that processing begins." So, as of November 2, the company has not published where data from the silent speech glasses goes or how long it is kept.
What the current policy does say, for the website and preorders:
- Stripe handles payment card details. Subvocal says it does not receive or store your full card number.
- Named service providers include Stripe, MongoDB, Netlify, Vercel, Cloudflare and PostHog. Gmail, Slack and Composio handle order messages.
- OpenAI is used “only when enabled to summarize aggregated internal analytics” for the team.
- The company says it does not sell personal information or share it for cross-context behavioral advertising.
On the product itself, the only statement is in the homepage FAQ. The company says it is "designing the system with privacy in mind" because typed text is "often highly personal or confidential." That is a stated intention. No product policy exists yet.
Workplace questions for IT teams
Silent speech glasses raise the same questions as any new input device, plus a few new ones.
Where does the text go? An input method can type into any app, including email and AI chat windows. Your existing AI and data rules still apply.
What leaves the device? Until Subvocal publishes a product privacy notice, you cannot answer this. Do not approve the glasses for confidential work until you can.
How does it pair with the computer? The company has not said. The answer affects your device management and software approval steps.
Is there a radio emitter on someone’s face? If the radar reports are accurate, the glasses emit radio signals. The company says delivery is subject to FCC approval. Some secure or clinical spaces have rules on radio devices. Check yours.
Is it a recording device? The company says there is no camera and input is intentional. That makes silent speech glasses simpler to allow than camera glasses. Our post on Meta smart glasses covers how camera models signal recording.
Accessibility: what the company has said
Subvocal’s homepage, preorder page and legal pages, as of November 2, do not describe accessibility uses. The company pitches its silent speech glasses for privacy and quiet in shared spaces.
We will not assume it suits people with speech or motor disabilities. The glasses still need deliberate mouth movement and a personal calibration step. If a disability services office is interested, ask the company directly whether it has tested with those users.
How silent speech glasses fit with other glasses we cover
Subvocal’s glasses are an input device. Most smart glasses today are output devices: speakers, cameras or a screen in the lens.
Our AR glasses explainer sorts glasses by what they show you. The HUD glasses vs media glasses comparison covers the two main display types. Subvocal fits neither group. As described, it has no display at all.
The closest comparison is the Meta Neural Band. As our Meta post explains, that wristband reads forearm muscle signals and turns finger movements into clicks. Subvocal applies a similar idea to words.
Subvocal’s coding-agent demo also fits the trend in our physical AI agent devices landscape: new hardware for giving instructions to AI. Both are early.
Questions to ask before buying silent speech glasses
If you or a client is considering a preorder, these are the questions to put to the company in writing:
- Where is silent speech decoded: on the glasses, on the computer or on a server?
- Is sensor data kept, and is it used to train models? Can a user or an organization opt out?
- Which operating systems and apps work at launch?
- What accuracy and speed did testers get, on what vocabulary, after how much calibration?
- What happens to the deposit if FCC approval is delayed?
- Is there an admin or managed-device option for organizations?
The terms say product details will come before final payment, and the deposit stays refundable until then.
Frequently Asked Questions
What are Subvocal’s silent speech glasses?
They are glasses from The Subvocal Company, a San Francisco startup, built to turn words you mouth without sound into text on your computer. The company calls them an input method for messages, notes and AI tools.
Do the Subvocal glasses have a camera?
No. Subvocal’s FAQ says its silent speech glasses use no cameras and no camera-based lip reading.
How do the glasses detect silent speech?
Subvocal’s own pages do not name the sensor. Digital Trends reports that, according to the company, the glasses send radar signals through the face and an AI model turns changes in those signals into text.
How fast are silent speech glasses from Subvocal?
Subvocal claims more than 200 words per minute. That figure has not been independently verified, and the company has not published test details.
How much do the Subvocal glasses cost?
The Early Bird price is $499 per pair, with a $199 refundable deposit and a $300 balance due later. Taxes, shipping and optional prescription lenses cost extra.
When will the glasses ship?
Subvocal estimates the first batch will ship in Q1 2027. Its terms say this is not a guaranteed date, and its preorder page says delivery is subject to FCC approval.
Can people outside the US order them?
Not yet. Paid preorders are open only for US delivery addresses.
Does the privacy policy cover data from the glasses?
No. The current policy covers the website and preorders. It says product data will get a separate or updated notice before that processing begins.
Can the glasses read my thoughts?
Subvocal says no. Its FAQ says the system only decodes words you intentionally articulate, and it is not always listening.
