TechnologyTuesday, November 3, 20261 min read

Subvocal raised $1.3M. A developer recreated it with AirPods in 9 minutes

By Hacker News · Published by Everything Blog

In short

The Subvox application connects to head-tracking headphones on macOS 14 or later to record acceleration, rotation rate, and attitude changes at roughly 25 Hz. Users generate session-specific motion templates by silently mouthing words, which the system then matches against new inputs using dynamic time warping. Because physical adjustments like re-seating the earbuds alter readings, the software verifies transcription signals through a 500-iteration label-permutation test requiring a p-value below 0.05.

Key points

Silent speech recognition can be prototyped using standard head-tracking earbuds instead…: Silent speech recognition can be prototyped using standard head-tracking earbuds instead of complex specialized hardware.
The system captures mechanical jaw and head movements rather than relying on vocal cord v…: The system captures mechanical jaw and head movements rather than relying on vocal cord vibrations or traditional language models.
Motion-based templates remain fragile because minor physical shifts like re-seating the A…: Motion-based templates remain fragile because minor physical shifts like re-seating the AirPods can distort the recorded signal.

Sources checked by our analysis pipeline: 1

A macOS prototype that tests whether AirPods Pro head-motion data (CMHeadphoneMotionManager

) carries enough information to turn silently mouthed words into text.

It doesn't use a language model. It's a small-vocabulary template recognizer plus a statistical test that tells you whether the signal is real.

brew install xcodegen # once

xcodegen generate

open Subvox.xcodeproj # ⌘R

Or from the command line:

xcodebuild -project Subvox.xcodeproj -scheme Subvox -derivedDataPath build -allowProvisioningUpdates build

open build/Build/Products/Debug/Subvox.app

xcodebuild -project Subvox.xcodeproj -scheme Subvox -derivedDataPath build test

Requires macOS 14+, AirPods Pro or another headphone that supports head tracking, connected as the Mac's audio output. On first launch, grant Motion. The mic and speech prompts only appear when you start the mic.

Screen | What it does |

---|---|

Sensors | Live plots of user acceleration, rotation rate and attitude, the measured sample rate, a mic input picker and level meter, and a recorder for experiment sessions. |

Train | Pick a word, hold Space, mouth it, release. Each take is stored as a template. Evaluate runs leave-one-out classification and a 500× label-permutation test, then shows accuracy…

AI Analysis

Tone

Neutral

In short

Key Points

Silent speech recognition can be prototyped using standard head-tracking earbuds instead…: Silent speech recognition can be prototyped using standard head-tracking earbuds instead of complex specialized hardware.
The system captures mechanical jaw and head movements rather than relying on vocal cord v…: The system captures mechanical jaw and head movements rather than relying on vocal cord vibrations or traditional language models.
Motion-based templates remain fragile because minor physical shifts like re-seating the A…: Motion-based templates remain fragile because minor physical shifts like re-seating the AirPods can distort the recorded signal.

Topics

Technology

Was this summary useful?

AI Analysis

Tone

Neutral

In short

Key Points

Silent speech recognition can be prototyped using standard head-tracking earbuds instead…: Silent speech recognition can be prototyped using standard head-tracking earbuds instead of complex specialized hardware.
The system captures mechanical jaw and head movements rather than relying on vocal cord v…: The system captures mechanical jaw and head movements rather than relying on vocal cord vibrations or traditional language models.
Motion-based templates remain fragile because minor physical shifts like re-seating the A…: Motion-based templates remain fragile because minor physical shifts like re-seating the AirPods can distort the recorded signal.

Topics

Technology

Was this summary useful?

Vercel CI no longer requires a linked Vercel project
Continuous integration setups now allow teams to bypass creating placeholder deployments simply to test code changes. Developers can activate automated testing workflows straight from their local terminal, provided their accounts are properly connected and hold the appropriate repository permissions. This change removes previous administrative friction by letting builds run immediately upon code updates.
1 min read
What's new in ClickStack - September and October '26
Observability teams gain broader dashboard visualization capabilities and system-level workload tracking across shared infrastructure in versions 2.39 through 2.42. The updates refine investigation workflows through experimental ad hoc multi-source filtering and an automated tool designed for agentic service health analysis. These changes improve operational visibility while modernizing collector telemetry alignments and query attribution.
19 min read
Building connections across the research infrastructure
Sixty-five representatives from 25 institutions gathered at MIT.nano to address shared laboratory challenges like equipment scheduling, billing, and interdisciplinary collaboration. Sponsored by the Northeast Microelectronics Coalition hub, the event centered on expanding open-source operational tools to help researchers seamlessly track and access specialized infrastructure. These collaborative efforts aim to replace outdated facility management processes and connect capabilities nationwide.
6 min read

Subvocal raised $1.3M. A developer recreated it with AirPods in 9 minutes

In short

Key points

More in Technology