· via Hacker News – Front Page (native)
Hex ships local voice dictation with a hold-a-key, paste-anywhere workflow
Hex, a voice dictation app that surfaced on Hacker News, transcribes speech locally and inserts it wherever your cursor is when you hold a key. Installers are available for macOS, Windows and Linux.

A new desktop tool called Hex wants to make dictation as frictionless as typing: hold down a key, speak, and your words land in whatever app you are using. The project reached Hacker News's front page on November 18, 2026, where it was presented as local voice dictation, meaning the transcription is meant to happen on the machine itself rather than through a cloud service.
How the workflow works
According to the Hex project site, the interaction model is deliberately minimal. On the Mac, you hold the Option key while speaking, and Hex then inserts the transcribed text into whatever field currently has focus. There is no separate window to position, no microphone toggle in a menu bar to hunt for, and no copying and pasting on your part — the app handles the hand-off into the active application itself.
That paste-anywhere approach is the core of the pitch. Rather than confining dictation to a dedicated notes surface, Hex treats speech as a system-wide input method that behaves the same way in an editor, a chat client, a browser field or a terminal. The site is terse, and it does not specify which speech recognition engine powers the transcription, what hardware requirements exist beyond the Mac build, or how the app is licensed and priced.
Platforms and installation
The Mac build targets Apple silicon machines running macOS 15 or later, and the site advertises Homebrew as the install route via the anomalyco/tap/hex cask. Windows and Linux are covered as well: Windows users get a PowerShell installation script, while Linux users can choose between a shell script installer and a downloadable .deb package.
The breadth of platform support at launch is notable for a tool in this category, where dictation utilities often start Mac-only. It suggests the developers are aiming at a general desktop audience rather than a single platform's ecosystem.
Why it matters
Two aspects of the launch stand out. The first is the local processing claim. Voice is among the most sensitive data types a computer handles, and dictation tools that route audio through remote servers create a privacy trade-off many users are unwilling to make, particularly for work involving confidential material. If Hex's transcription genuinely stays on the device, that concern largely disappears, and the tool becomes viable in contexts where cloud dictation is a non-starter.
The second is the workflow design. Operating systems ship with built-in dictation, and plenty of apps have voice input of their own, but the experience is usually inconsistent from app to app and easy to forget exists. A single held key that works everywhere lowers the cognitive cost of switching between typing and speaking, which is the kind of small ergonomic change that can turn dictation from an occasional accessibility feature into an everyday input habit.
The interest on Hacker News's front page indicates there is an appetite among technical users for exactly this combination: local-first processing, a keyboard-centric trigger, and behavior that is uniform across the whole desktop. What the project site does not yet answer is how accurate the recognition is in practice, how it performs across accents and languages, and what the sustained development model looks like. Those questions will determine whether Hex is a lasting utility or a well-executed first release.
- #voice-dictation
- #desktop-apps
- #macos
- #windows
- #linux
- #productivity