YC F26 Startup Subvocal: Silent Mouthing to Text at 200+ WPM
The Subvocal Company, backed by YC F26 and Afore with ~$1.3M, opens US preorders for camera-free glasses that turn deliberate silent mouthing into text.
Dictating email in a café is often more awkward than productive. The Subvocal Company, in YC's Fall 2026 batch, wants a quieter path: smart glasses that turn deliberate, near-silent mouthing into text—without speaking aloud or touching a keyboard.
The startup says it has raised about $1.3 million from Y Combinator, Afore Capital, and angels. The product site stresses no cameras: no video lip-reading and no laptop webcam. Sensing is contactless physiological pickup around the orofacial muscles; coverage describes radar-style reflections, while YC job posts keep the broader design space (RF, EMG, EEG, mmWave and related methods) undisclosed for IP reasons. An AI model maps those micro-movements to text. The company claims more than 200 words per minute—speaking speed without sound.
US preorders are open: a refundable $199 deposit locks the $499 Early Bird price (marked down from $699), with the balance due later. Shipping is estimated for Q1 2027; prescription lenses will cost extra. Intended uses include messaging, notes, writing, and talking to AI assistants—plus the more cinematic pitch of two wearers holding a private conversation across a crowded room. For people already using tools like Wispr Flow, Subvocal is not selling better speech recognition. It is selling the right not to speak at all.
The hardware is still in development. The 200+ wpm figure has not been independently verified, and accents, rare words, and long silent dictation will decide whether demos survive daily use. Cross-person generalization will likely need a short calibration. Even so, the bottleneck is real: typing is slow, public dictation is socially expensive, and a silent high-speed channel is worth probing with real sensors.
If the first pairs ship next spring and actually turn mouthed drafts into email in a quiet office, Subvocal will stop looking like another glasses concept reel—and start rewriting the default posture of AI input.