New Delhi: A technology startup founded by two young entrepreneurs, Aarush Yadav and Arnav Kumar, is developing smart glasses that can convert silently mouthed words into digital text. The wearable, created by The Subvocal Company, aims to let people write messages, draft emails and interact with artificial intelligence (AI) assistants without speaking aloud or using a keyboard.

The startup has announced that it has raised $1.3 million from Y Combinator and Afore Capital to develop the technology. Its founders say the glasses can decode subtle, almost invisible mouth movements into text at speeds exceeding 200 words per minute.

The concept offers a different approach to voice dictation. Instead of recording speech and converting it into text, the device is designed to detect the small movements associated with forming words silently.

How do Subvocal’s AI glasses work?

Conventional voice-to-text systems rely on microphones to capture spoken words. Camera-based lip-reading systems, meanwhile, analyse visible mouth movements. Subvocal says its approach does not require either a microphone or a camera for silent speech input.

The company describes its technology as non-contact sensing. When a person deliberately forms words without producing audible speech, tiny movements in the face and muscles involved in speaking create signals that the wearable is designed to detect. An AI model then interprets those signals and converts them into text.

Public reports describe the sensing approach as using radio-frequency or radar signals. These signals are understood to change as they interact with subtle movements around the user’s mouth. The company, however, has not publicly disclosed its complete sensing architecture, citing competitive considerations.

The technology does not read a person’s thoughts. Users must still consciously form words with their lips. The difference is that they do not need to make an audible sound while doing so.

According to the startup, the movements can be subtle enough to remain difficult to notice from a normal conversational distance. The glasses are intended to provide a discreet way of entering text in places where speaking to a computer may be inconvenient.

What can users do with the glasses?

Subvocal envisages the wearable as an alternative input method for everyday computer and AI interactions. Potential uses include:

  • Writing messages: Users could compose and enter messages without typing on a phone or computer.
  • Drafting emails: Silent mouthing could be converted into text while working in an office, travelling or sitting in a public place.
  • Taking notes: The device could help users record thoughts and notes without speaking aloud.
  • Interacting with AI assistants: Users could enter prompts and instructions for AI tools without using a keyboard or microphone.

The technology may be particularly useful in environments such as libraries, shared offices, public transport and cafes, where speaking aloud to a device can be uncomfortable or disruptive.

The startup has also suggested that the technology could eventually support silent exchanges between two people using compatible devices. However, this remains a potential application rather than an established everyday feature.

Startup raises $1.3 million

Co-founders Aarush Yadav and Arnav Kumar are developing the product through The Subvocal Company. The reported $1.3 million funding round includes backing from startup accelerator Y Combinator and venture capital firm Afore Capital.

In announcing the funding, Yadav said the product was designed to address two limitations of conventional computer interaction: typing can be slow, while speaking to a computer can feel awkward in public.

The company says its prototype can interpret silent mouthing at more than 200 words per minute. That figure is a central part of its pitch, as it would place the proposed input method closer to the speed of ordinary speech than conventional keyboard typing.

However, the claimed performance should be treated with caution. The speed figure comes from the company’s own prototype claims and has not been independently verified through publicly available testing.

Price, reservations and expected availability

The Subvocal Company has opened reservations for the glasses in the United States. The listed price is $699, while an early-bird offer puts the price at $499.

Customers can reserve a pair with a refundable deposit of $199, with the remaining $300 payable later under the company’s reservation arrangement. Prescription lenses are expected to cost extra.

The startup currently estimates that shipments will begin in the first quarter of 2027. The timeline is not guaranteed, as the product remains under development. The company has also indicated that final specifications may change and that deliveries are subject to regulatory approval.

The current reservation offer is for US customers; the company has not announced a confirmed launch date or pricing for India.

Accuracy and real-world performance remain key questions

While the idea has attracted attention, the technology will need to demonstrate that it works reliably beyond controlled demonstrations. Its ability to interpret different accents, unfamiliar words, varied speaking styles and long stretches of silent input will be important to its practical success.

Accuracy is particularly significant because users will not be producing audible speech that can be checked against a conventional recording. The system must interpret subtle physical signals and distinguish intended words consistently.

Comfort, battery life, calibration requirements and performance across different users will also influence whether the glasses can become a practical everyday device.

For now, Subvocal represents an emerging approach to human-computer interaction rather than a proven replacement for keyboards or voice assistants. If the company can deliver reliable performance at scale, its technology could offer another way to communicate with computers and AI tools without speaking aloud.