what is saysomething fr?

Say Something is a free, open source voice dictation app for Windows and Mac that lets you talk instead of type. You hold a key, Right Ctrl on Windows or Right Cmd on Mac, say what you want, let go, and the cleaned up text appears at your cursor about a second later. Everything runs on your own computer using an on device version of Whisper for speech recognition, so there is no cloud service, no account, no subscription, and no telemetry. One feature the README highlights is the drop pad. Instead of always typing into whatever window currently has focus, you can hold a different key, Right Alt on Windows or Right Option on Mac, and your words land in a small floating pad that is already copied to your clipboard. You can then drag that pad onto the exact text field you want and let go, or just paste with Ctrl+V anywhere. There is also a hands free mode where a quick tap starts continuous listening that stops automatically once you go quiet, with an adjustable silence window. An optional AI cleanup feature can polish your dictated text using a local Ollama model, offering styles like Clean up, Professional, Casual, or Bullet points. This is turned off by default, only talks to Ollama on your own machine, and if the local model is slow or fails you simply get your original unedited text back. Other features include smart formatting that removes filler words like um and uh, spoken commands such as saying new line or new paragraph, a custom dictionary for names and jargon, and a local history of your last 200 transcriptions stored as a plain JSON file. The README compares the project to Wispr Flow, a paid cloud based competitor, noting that Say Something is free and works offline, while Wispr Flow streams text live into fields as you talk, something Say Something does not do yet. Setup requires Node.js 24 or newer on both platforms, plus the Windows C# compiler or the Xcode Command Line Tools on Mac to build a small local helper. After npm install and a setup script that downloads the default speech model, npm start launches the app into your system tray or menu bar. The project is released under the MIT license.