The Subvocal Company
Murmur — Control your computer in silence.
under-chin wearable
Control your computer in silence.
Work without typing or speaking.
An under-chin wearable that turns silent speech into commands and text.
Send messages mid-meeting, take notes while walking, and run tools hands-free.
How it works
Step 01 — You mouth the words. Subvocalize naturally; our sensor reads the muscle signals your jaw, tongue, and throat produce.
Step 02 — We decode the signal. Real-time biosignal decoding converts muscle patterns into text or commands, expanding over time.
Step 03 — Your apps execute. Send a message, create a note, or steer an AI agent — every action previewed before it fires.
Interactive demo
See what the sensor sees. No sound. No visible talking. Just under-chin muscle signals.
01 Capture: Speech muscle signals, measured under your chin.
02 Segment: Word boundaries detected in the signal.
03 Decode: Each word inferred with live confidence.
04 Execute: Your tools respond. Silent commands, real actions.
Built for privacy and control.
No audio. No camera. Actions only happen with clear intent and clear receipts.
No microphone. No camera. The sensor reads muscle signals under your chin, not your surroundings. No ambient audio capture, no video or lip reading. Signal data is treated as sensitive.
Intentional activation: You decide when Murmur senses for commands. Hold-to-speak mode, quick pause and resume, clear active indicator.
Previews and undo: Risky actions require confirmation. Everything is reversible. Preview before send, confirm for high-impact actions, instant undo with a receipt.
Immediate utility, not a science project.
Start with a reliable command set. Expand as you go.
Messaging: Reply in Slack, Teams, or email drafts; short messages while walking; quick reactions and canned responses.
Notes: Capture a note instantly; create action items; add tags and reminders.
Control: Scroll, click, switch tabs; search and open apps; run a saved workflow.
Discreet under-chin wearable.
A disposable skin patch and a reusable puck. Designed to disappear.
Patch and puck: Disposable adhesive patch. Reusable sensor puck.
Lightweight and low profile: Sits under the chin, not on the face.
All-day friendly: Built for comfort and repeatable placement.
Audio-first interface: Feedback through open-ear audio, not a screen.
Quick calibration.
Everyone's signal is different. We personalize to you.
01 First fit: Apply the patch and attach the sensor puck.
02 Say a few phrases internally: The system learns your unique signal patterns.
03 Start with a reliable command set: Begin with core commands and expand over time.
Works where you work.
Starts with everyday tools. Expands via workflows. Platforms: macOS (live), Windows (planned), Web (planned). Apps: Slack, Gmail, Calendar, Notes (live), Browser (planned).
Expand safely over time.
Now: Reliable command set and confirmations.
Next: Larger vocabulary and better cross-session stability.
Later: Faster dictation and deeper workflows.
Common questions.
Is this recording audio?
No. There is no microphone. The sensor reads electrical signals from the muscles under your chin. No sound is captured, stored, or transmitted.
Is there any camera or lip reading?
No camera, no video, no lip reading. The entire system works from surface muscle signals. Your surroundings are never observed.
Do I have to move my mouth?
Barely. You subvocalize — engaging your speech muscles at a level that produces no visible talking and no obvious mouth movement. It feels like speaking internally.
How does it avoid false triggers?
Commands require intentional activation. Hold-to-speak mode or explicit on/off toggle. Casual jaw movements, chewing, or swallowing are filtered out by the signal model.
What does calibration look like?
A short guided session where you repeat a handful of phrases silently. Typically about two minutes. The system adapts to your unique signal patterns and improves over sessions.
What devices and apps are supported?
We start with macOS and integrate with Slack, Gmail, Calendar, and Notes. Windows and web support are on the roadmap. You can extend functionality through saved workflows.
What happens if it mishears?
Every action is previewed before it fires. High-impact actions like sending an email require explicit confirmation. Everything comes with instant undo and a clear receipt.
What does early access include?
A developer unit (sensor puck and adhesive patches), access to the desktop app, and a direct feedback channel with our team. We invite in small waves to ensure quality.
Get early access.
Join the waitlist for the first builder units and private beta. We invite in small waves.