- Who it helps
- Mac users and accessibility researchers evaluating alternatives to spoken dictation.
- Availability
- Public proof-of-concept repository checked November 1. Requires macOS 14+, camera and Accessibility permissions; Apple Silicon is recommended.
What the prototype adds
The menu-bar app captures mouth and neck movement, decodes a phrase and inserts text. Personal adapters can be trained from recorded prompts. An optional local language model cleans up the result; fluent wording can still conceal a recognition error.
Check the model terms
The application code is MIT-licensed. Its README assigns separate research-use terms to downloaded pretrained weights. It also warns that silent webcam speech is harder than the voiced training benchmark. These limits matter before considering deployment.
Hushkey: a local silent-dictation prototype ↗Chaplin implementation ↗
What you can do
- Read the limitations and weight terms before evaluating the prototype. Open reference ↗
Where the evidence stops
No installation, camera capture or inference was performed. Accuracy, accessibility suitability and hardware performance remain unverified.
Sources read. Sources checked November 1, 2026 UTC. No installation, inference or end-to-end workflow was executed.
Sources & further reading
GitHub · official
Hushkey: a local silent-dictation prototype ↗
Identifies the subject, documented capability or specific verification limit.
github.com · page checked 2026-11-01
GitHub · implementation
Chaplin implementation ↗
Identifies the subject, documented capability or specific verification limit.
amanvirparhar · page checked 2026-11-01
Coverage note: Independent attention and end-to-end execution remain unverified.
The record over time (1)
Dated source checks and published explanations. An observation date can differ from the original release date.
- · Added to Current
Hushkey turns silent mouth movements into text at the Mac cursor, using a webcam and a local speech-reading model.