All stories
Subvocal Launches Community reports

Subvocal wants you to talk to AI without making a sound

The founders' RF-sensing glasses decode deliberate silent articulation into text. Preorders require a $199 deposit, with delivery estimated for Q1 2027 and subject to FCC approval.

By AITrending2 min read
Sources (4)

Why now The founders introduced the prototype and announced preorders in an English Show HN post on November 2 at 17:16:17 UTC. The live preorder page supplies pricing and delivery conditions.
Subvocal's official demonstration shows two seated presenters, one wearing wired prototype glasses, in front of a screen labeled Typing.
Frame from the company's English prototype demonstration, showing experimental glasses and connected hardware. This is company-published footage, not the final shipping design or an independent speed test. Source: The Subvocal Company

Dictating to a coding agent in a quiet office is the problem Subvocal wants to solve with glasses, rather than a microphone. The Subvocal Company introduced its silent-input prototype and announced preorders on November 2. Its founders say wearers can deliberately articulate words without speaking aloud and have those movements decoded into text.

The company markets input at more than 200 words per minute and shows a prototype demonstration. That is an advertised capability, not an AITrending measurement of speed, accuracy or performance across users. The founders say they are still doing research and engineering to turn the prototype into the product depicted on the website.

Radio sensing, not thought reading

In their English Show HN post, founders Aarush and Arnav describe switching from facial electrodes to radio-frequency sensing. Antennas in the frames send signals through the face; silent articulation changes the signals as they interact with the tongue, lips and soft palate. Models trained on recordings from people wearing the glasses decode those changes into text, according to the founders.

The product page explicitly says it does not read arbitrary or involuntary thoughts. It requires intentional articulation. The founders describe a short calibration in their linked demonstration, rather than claiming identical performance without setup for everyone.

The intended workflow is text input for existing computer applications, including conversations with AI and instructions to coding agents. The reviewed pages do not establish a final operating-system compatibility list, an installation procedure or whether every supported configuration processes signals locally. Silent input should not be confused with a verified guarantee that data never leaves the device.

A preorder, with delivery still ahead

The live preorder page lists a $199 deposit toward a $499 early-bird total, leaving $300 due later. Taxes, shipping and optional prescription lenses cost extra. It says the deposit is refundable before full payment.

Estimated shipping is Q1 2027, and delivery is subject to FCC approval. Those conditions make this an early hardware bet for people who want voice-like input around coworkers or in quiet rooms. They do not establish that the final glasses have shipped, received approval or reached the advertised speed in independent testing.

Related reading