Subvocal Glasses Turn Silent Mouthing Into Text at 200 WPM

Subvocal glasses are the first product from The Subvocal Company, a San Francisco startup that says it has raised $1.3 million from Y Combinator and Afore Capital. Founder and CEO Aarush Yadav announced the round on X on October 30, saying the glasses decode silent, almost invisible mouthing into text at more than 200 words per minute.

We’ve raised $1.3M from YC & Afore Capital so you never have to type or speak again. Typing is slow. Talking into your computer is awkward. Subvocal fixes both.

— Aarush Yadav (@AarushYadav47) October 30, 2026

The post has passed 11 million views.

How Subvocal Glasses Work

According to Yadav, the glasses track the muscles used for internal speech with non-contact sensing. The company says there are no cameras involved and no lip reading through a webcam. You mouth words as subtly as you like, and at normal conversational distance the movement is designed to be invisible.

Subvocal also stresses that it does not read thoughts. It only decodes words you deliberately articulate, so it is a very different idea from the brain implant approach behind Neuralink’s Telepathy project. Setup includes a short personal calibration, and the startup admits that current prototypes still make mistakes, with accuracy improving as it collects more data.

The pitch is simple: voice dictation is fast but awkward on a train or in an open office. Subvocal is being built as an input method rather than an app, so the text should land in ChatGPT, Claude, Cursor, Slack, email or anywhere else you would normally type.

Price, Preorders and Shipping

The Subvocal preorder page lists an Early Bird price of $499 per pair, down from $699. Buyers pay a $199 refundable deposit now and the remaining $300 later. Paid preorders are open only in the United States, prescription lenses cost extra, and delivery depends on FCC approval.

First shipments are estimated for Q1 2027, but the company says the product is still in development and final specifications are not set. It has not published accuracy figures or preorder numbers.

Developers Are Already Copying the Idea

The launch quickly drew imitators. Developer Joe Blau said he recreated the concept in nine minutes for AirPods Pro, and his open source Subvox prototype tests whether AirPods head motion data can turn silently mouthed words into text. It uses a small vocabulary template matcher rather than a language model, so it is an experiment, not a rival product.

Subvocal enters a crowded face computer race. Meta keeps pushing its Ray-Ban line despite issues such as a recent mandatory Meta glasses security update, and newcomers like Nothing are lining up too, with the company reportedly planning to enter the smart glasses market by 2027. None of them offers silent typing yet, which is the gap Subvocal is betting on.