23 Daily — According to news.ycombinator.com, Aarush and Arnav, two students from Georgia Tech, launched the Subvocal smart glasses in 2026. The device enables users to communicate silently with computers by precisely decoding lip movements and internal speech actions via radio frequency (RF) signals, converting silently mouthed words into text at speech speed.
Radio Frequency Technology Enables Silent and Fast Input
Initially, the Subvocal team experimented with surface electromyography (EMG) sensors to detect users' speech muscles, but instability led them to adopt RF technology. RF signals penetrate muscles and skin, capturing detailed internal mouth movements. The system can recognize internal speech motions without opening the mouth. The team noted users do not need to open their mouths; mouthing words with lips closed works after calibration.
Project Origins and Technical Challenges
Aarush and Arnav began by attempting to replicate MIT Media Lab's AlterEgo silent speech system within 36 hours at Stanford's TreeHacks hackathon, using EMG electrodes on the face. Although the prototype had limited letter-level accuracy and was rudimentary, it laid the groundwork for their later more stable and effective RF system.
Editor's take: Opportunities in Privacy-Respecting Communication Technologies
Subvocal smart glasses represent a novel communication mode, suitable for quiet environments like libraries and offices, or situations where vocal input is impractical. Unlike conventional voice recognition, it offers high privacy and efficient silent input. Entrepreneurs exploring similar innovations should focus on the stability of RF sensing, the collection of mouth movement data for accuracy, and real-world testing to verify response speed and usability in daily contexts.