Hex: local voice dictation for macOS, Windows and Linux
Hex is a MIT-licensed, Rust-built dictation app for macOS, Windows and Linux that keeps speech recognition entirely on your machine, with 3,342 GitHub stars.
TL;DR: Hex is a speech-to-text app that runs entirely on your own machine. The repository description is short, “Local voice dictation for mac, windows and linux”, and the GitHub topics back it up with
local-first,dictation,macos,linuxandrust. It is MIT licensed, was created in July 2026, and took about 1,200 stars in today’s trending window despite being only a few months old.
What it is
Hex is a desktop dictation tool. You speak, it turns that into text, and the text lands wherever you are typing. The word doing the work in the description is “local”. Nothing about the project suggests a server component, an account, or a per-word subscription.
The topic tags describe the design rather than the marketing: local-first sits next to dictation and rust, and the platform tags are macos and linux. The description names all three desktop platforms, so Windows is covered by the project even though it does not carry its own topic tag.
This is a voice-to-text tool, not a voice assistant. The difference matters. Hex is not going to open your calendar, summarize your day or answer questions. It is a faster way to produce text.
Why it matters that it is young
The repository was created on 2026-07-17. That is roughly three and a half months before this post. It already has 3,342 stars and 192 forks.
Growth like that in a short window usually means one of two things: a genuinely good product, or a strong launch. The default branch is named dev rather than main, which tells you the project is still in active development and that the default branch is the one receiving the work.
It also means the API surface, the configuration and the release process are all likely to move. If you are evaluating Hex as a daily driver, pin a release rather than tracking the default branch.
Local-first is a checkable claim
The best way to evaluate a local-first tool is to watch the network. Before you dictate anything sensitive, check that the process is not talking to a remote host.
# Watch outbound connections while you dictate
sudo tcpdump -i any -n 'host not 127.0.0.1 and host not ::1'
Or inspect the sockets the application currently holds:
# List established connections for the process
ss -tunp | grep -i hex
If the claim holds, you will see nothing during dictation. If you do see traffic, the “local” in the description is doing less work than it appears to.
How to try it
The repository is the starting point:
git clone https://github.com/anomalyco/hex.git
cd hex
The project homepage listed in the metadata is anoma.ly/hex, which is where release builds for each platform are published. Read the install instructions there before building from source, because a Rust desktop app on three operating systems is rarely a plain cargo run away from working.
When to use it
Hex makes sense if you type a lot, if you would rather not send voice recordings to a cloud vendor, and if you are on one of the three desktop platforms it names.
It is a weaker fit if you dictate on mobile, since mobile is not in the description. It is also a weaker fit if you need meeting transcription with speaker labels, because nothing in the metadata suggests that exists.
Source and Accuracy Notes
- Project page: https://github.com/anomalyco/hex
- Source repository: anomalyco/hex (verified via GitHub API
html_url) - License: MIT (verified via GitHub API
license.spdx_id) - Stars at time of writing: 3342
- Homepage listed in repository metadata: https://anoma.ly/hex
- Repository created: 2026-07-17, default branch
dev
Related Posts
ai-setup
Recall – Persistent Memory for Claude Code via MCP Hooks
Recall gives Claude Code a permanent memory store that survives session restarts and context compaction. Four hooks capture and restore context automatically — with cloud SaaS or self-hosted options.
2/28/2026
dev-tools
Automotive Skills Suite for AI Engineering
Evaluate Automotive Skills Suite for APQP, ASPICE, HARA, safety-plan, and DIA workflows with setup notes, governance risks, and SME review guidance.
5/28/2026
dev-tools
awesome-agentic-ai-zh Roadmap Guide
Explore awesome-agentic-ai-zh as a Chinese agentic AI learning roadmap, with setup notes, track selection, study workflow, and evaluation guidance.
5/28/2026