Beating AI News Flash: YC-incubated startup Subvocal has announced cumulative funding of $1.3 million and is developing a pair of "silent typing" glasses. Users simply mouth the words without speaking or touching a keyboard, and text can be input into a computer. Writing emails, taking notes, or even giving commands to an AI Agent can all be done this way. The team claims the prototype can continuously transcribe over 200 words per minute.
The glasses use low-power radio frequency signals to capture subtle movements of facial speech muscles, which are then decoded by an AI model. Users still have to move their mouths, but do not need to make sound.
Similar technology is not new. MIT introduced AlterEgo back in 2018, which used electrodes attached to the face to recognize silent speech, but at the time it was mainly tested on a limited vocabulary of about 20 words. Subvocal aims to go further and achieve contact-free, vocabulary-unrestricted continuous input.
However, the product is still some distance from actually hitting the market. The current demo prototype still drags around an external circuit board and looks quite different from the planned glasses form factor. The speed of over 200 words per minute has also not been verified by public independent evaluation.
Subvocal has opened pre-orders in the United States, with an early-bird total price of $499 and a refundable $199 deposit required upfront. The company expects to begin shipping in the first quarter of 2027, subject to FCC approval.
