The Subvocal Company's Wearable Aims for 200 Words Per Minute in Silence

The Subvocal Company is not trying to build a better keyboard. Its founders, Aarush Yadav and Arnav Kumar, are betting that the most efficient way to input text might be to not move your hands at all. Instead, their wearable device, called Murmur, reads the electrical signals from the muscles in your throat as you silently form words, translating deliberate subvocal speech into computer commands and text [Y Combinator, Oct 2026].

The Hardware Wedge

The product takes two public forms: a pair of glasses and a discreet under-chin patch. Both use muscle-sensing electrodes to capture biosignals, bypassing microphones and cameras entirely [Dealroom, Sep 2026]. The initial target is the knowledge worker's desktop, with integrations for macOS applications like Slack, Gmail, Calendar, and Notes [Founderland, Oct 2026]. The technical claim underpinning the ambition is speed. Public descriptions state the prototype can decode speech at rates above 200 words per minute, a figure that would rival or exceed proficient typing [Advance Intelligence, Oct 2026]. For developers, writers, or anyone who thinks faster than they can type, that's the core proposition: hands-free, silent input at conversational pace.

The Team and the Capital

The founders are Georgia Tech computer-science students who built an early prototype at Stanford's TreeHacks hackathon in February 2026 [Dealroom, Sep 2026]. Both have prior startup experience. Yadav previously founded GreenSat, an agri-tech startup, while Kumar co-founded Leap, an edtech and fintech platform for Indian students, and Gozoomo, a peer-to-peer used car marketplace [TechCrunch, Sep 2021][Forbes, Dec 2015]. They have assembled $1.3 million in pre-seed capital across three tranches, with lead checks from Afore Capital and Y Combinator, which accepted the company into its Fall 2026 batch [Advance Intelligence, Oct 2026]. The funding is earmarked for building out the core team, with open roles for a Head of Hardware, Head of Sensing, and Head of ML, all based in San Francisco [Y Combinator, Oct 2026].

April 2026 | 0.5 | M USD
September 2026 | 0.5 | M USD
September 2026 | 0.3 | M USD

The Scale-Up Challenge

The technical breakdown reveals where the real work begins. Decoding subvocal speech is a signal-processing problem with high variability between users. The system must filter out noise from swallowing, head movement, and ambient electrical interference to isolate the clean neuromuscular signature of intended speech. Then it must map that signal to text with high accuracy, a task that requires a personalized model for each user. Achieving 200 WPM in a controlled demo is one thing; maintaining that speed and 99%+ accuracy across a thousand different throats in noisy environments is another. The company's public materials do not yet detail the latency between thought and text, a critical metric for usability, or the planned onboarding calibration process.

The sober assessment is that this is a classic hard-tech gauntlet. The risks are not conceptual but executional, layered across hardware, software, and human biology.

• Hardware reliability. The wearable must be comfortable for all-day use, with sensors that maintain consistent skin contact without irritation. Battery life and form factor are non-negotiable for consumer adoption.
• Model generalization. The biosignal decoding model must work reliably across a diverse user base without requiring weeks of training data from each individual. Accuracy degradation for users with different accents, speech patterns, or anatomies could limit the market.
• Go-to-market timing. As a pre-revenue company accepting reservations, the path to a shippable product is long [Digital Trends, Oct 2026]. The space is currently uncontested, but that vacuum invites competition from well-funded AR/VR players or neural interface startups once the use case is validated.

The Subvocal Company's bet is that silent speech is a superior interface waiting for the right hardware. If the team can navigate the gauntlet from prototype to reliable product, they won't just be selling a faster input method. They'll be selling a new, private channel between the brain and the machine.

Sources

1. [Y Combinator, October 2026] Launch YC: The Subvocal Company: Talk without talking | https://www.ycombinator.com/launches/Wf9-the-subvocal-company-talk-without-talking
2. [Dealroom, September 2026] Subvocal raises $1.3M from Y Combinator to decode silent speech… | https://app.dealroom.co/news/feed/subvocal-raises-1-3m-from-y-combinator-to-decode-silent-speech-into-computer-commands
3. [Founderland, October 2026] Subvocal launches under-chin wearable for silent computer… | https://www.founderland.ai/articles/subvocal-launches-under-chin-wearable-for-silent-computer-co-muudsais
4. [Advance Intelligence, October 2026] Subvocal Says It Raised 1.3 Million Dollars From YC and Afore Capital for Silent Speech Glasses | https://www.advanceintelligence.ai/story/2026-10-30-12
5. [TechCrunch, September 2021] Owl Ventures and Harvard Management Company back India's study-abroad platform Leap in $55 million round | https://techcrunch.com/2021/09/07/leap-finance-raises-55-million-to-help-indian-students-study-abroad-plans-international-expansion/
6. [Forbes, December 2015] Gozoomo Gets $5M To Make Buying Used Cars In India Easier | https://techcrunch.com/2015/07/02/gozoomo/
7. [Digital Trends, October 2026] Subvocal's glasses promise to let you talk to your computer in complete silence | https://www.digitaltrends.com/computing/subvocals-glasses-promise-to-let-you-talk-to-your-computer-in-complete-silence/