Subvocal Develops Smart Glasses for Silent Speech-to-Text Interaction
New wearable technology translates silent mouth movements into digital text using radio waves.
1Silent Speech Interface Innovation
Subvocal is developing smart glasses that convert silent mouth movements into text, allowing users to communicate with computers without vocalizing. This technology aims to replace traditional keyboards and voice commands for private, high-speed input. Founders Aarush Yadav and Arnav Kumar believe this interface could redefine how professionals interact with digital systems.
2Technical Mechanism and RF Sensing
The device functions by projecting low-powered radio waves toward the wearer's face. These waves are distorted by muscle movements and facial shape changes during speech. The system captures these distortions to decode open-ended speech at speeds exceeding 200 words per minute, maintaining a low error rate despite the small dataset scale.
3Evolution from EMG to RF
The project originated at a hackathon using EMG electrodes attached to the face. Recognizing that sticking electrodes to the skin was not commercially viable, the team pivoted to RF sensing. This transition required building an extremely low-noise RF stack and optimizing antenna placement to ensure maximum field overlap with facial articulators.
4Rapid Development and Funding
Founded by Georgia Tech students, the startup secured initial funding from Afore Capital after a successful hackathon demonstration. Y Combinator partner Dave Lieb also provided funding after testing a prototype. The team has progressed from a table of taped electronics to a semi-portable board and functional glasses in just four months.
5Calibration and User Experience
To minimize setup time, the glasses utilize pre-trained AI models for speech recognition. During testing, users like Dave Lieb only needed to read 20 sentences to calibrate the headset to their specific speech patterns. This calibration process is essential for the device to accurately interpret individual facial movements into digital text.
6Design and Commercial Availability
The current demo headset is 3D printed and lacks a final aesthetic, but the company plans for a much sleeker design. Preorders are open for a Q1 2027 release, with an early bird price of $499. Customers can secure a unit now for a $199 deposit, with the remaining balance due upon shipping.
7Proof of Concept Limitations
While the core signal decoding is functional, the device remains a proof of concept. Key performance metrics such as battery life, long-term accuracy rates, and potential app integration details remain unspecified. The founders are currently focused on proving that the signal exists and can be reliably decoded for practical computer interaction.
8Future of Silent Productivity
Subvocal targets a future where emails, code, and messages are generated at the speed of speech without disturbing others. Whether this technology scales successfully or faces hardware challenges remains to be seen. If successful, it could usher in a new era of silent, high-productivity computing for workers across various professional industries.