Handy and Talkey solve the same basic problem: speak, then get text without sending your audio to a cloud service. Handy is the more complete app: install it, open its settings, choose a model, and dictate straight into the active text field. It supports Windows, macOS, and Linux, with Whisper and Parakeet models and configurable hold-or-toggle shortcuts.
Talkey is a smaller collection of scripts and a background service. It uses Python and faster-whisper, with AutoHotkey on Windows and a shell script on Linux. The daemon keeps the model ready while it’s being used and can unload it after idle time. The defaults give Turkish and English their own shortcuts—F8 and F9—and put the result on the clipboard. The keys and languages can be changed in its config.
The reason I made my own was Wayland. I wanted dictation without relying on simulated keystrokes to enter text into another app. Handy’s current Linux notes document extra input tools for reliable text entry—wtype or dotool, and ydotool for some Ubuntu setups. Talkey stops at the clipboard, so I can review the words and paste them where I want.
The trade-off
Handy is easier to install and use, works on macOS, and offers more model and GUI options. Talkey has a narrower workflow that avoids the text-injection step—but I have to paste manually and configure more things myself.
For someone who wants a ready-to-use app, direct insertion, and macOS support, Handy is likely the better fit. I built Talkey for a specific friction I had with Linux text input. It isn’t a replacement for Handy; it’s a smaller tool shaped around a different constraint.