Handy, an offline dictation app, climbs Trendshift at 35,800 stars

A paid dictation app sends your voice to its own servers. Handy does not. Handy keeps the audio on your own computer, and more people found it this week.

Handy is a free, open-source speech-to-text app built by CJ Pais. It is ranked 23rd on Trendshift's daily list on October 31. The GitHub repository holds 35,769 stars and 3,395 forks today. The repository was created on February 13, 2025. The code is MIT-licensed and written in Rust using Tauri v2.

What it does

Handy sits in the system tray. You press a keyboard shortcut, speak, and release. The app transcribes what you said. It pastes the text into whatever field you are typing in. The default mode is push-to-talk. A toggle mode is also available: press once to start, press again to stop. The key binding is configurable. In the maintainer's own demo on the project site, the shortcut is Control-Z.

The project page lists four claims about the app. It is free. It is open source. It is private, which means the audio stays on your own computer. It is simple: one tool, one job. The app runs on Apple Silicon Macs, Windows, and Linux.

The traction

Trendshift reports the rank, not the number of stars gained today. The star total shown here is the whole lifetime of the repository, which is almost nine months old. There are no release notes between those stars and this ranking to point at. The most recent release, v0.9.8, was tagged on October 3, 2026. The release notes read "bug fixes." Two earlier releases are described the same way.

So the news is not a new feature. The news is that developers are picking up this project right now. Trendshift noticed. Handy sits beside Wispr Flow, superwhisper, Otter and other paid dictation apps that charge a monthly fee. All of them send audio off the device by default. Handy's whole pitch is that it does not.

What a reader does with it

For somebody paying for a dictation subscription, Handy is an open-source alternative. It answers two questions people often ask of these tools. Where does my voice go? Who pays for the subscription? Here the answers are nowhere, and nobody.

For a team adding accessibility features, Handy is an existing MIT-licensed project. The team can study it, borrow from it, or depend on it. The repository's listed topics are accessibility, cross-platform, speech-to-text, and tauri-v2. The project page says "accessibility tooling belongs in everyone's hands, not behind a paywall."

For anyone evaluating speech-to-text models, Handy already does the push-to-talk capture. It handles the paste integration and the system tray plumbing. The model is swappable.

For a reader whose dictation app sends every word to a server, Handy keeps it local. The install is a short list of trade-offs that fit in one app.

Source

cjpais/Handy on GitHub, and the project site at handy.computer.