Hey Product Hunt! 👋 I'm the maker of Sonora.

Honest origin story: I wanted a voice-dictation app like Speakly or Wispr Flow, the kind where you press a key, talk, and the text lands right where your cursor is. Then I hit the usual wall: not available on Linux. Like a lot of great tools, it's macOS/Windows only.

So I built the thing I wished existed. 🎙️

Sonora is an open-source, Spotlight-style floating dictation bar for macOS, Windows AND Linux. Hit a global shortcut, speak, and the text shows up live, either typed straight at your cursor or copied to your clipboard.

What makes it different:

- 🐧 Actually runs on Linux (Wayland + X11), not just mac/Windows

- 🔌 Bring your own engine: Gemini (live streaming), Mistral, OpenAI, Groq, or any OpenAI-compatible endpoint

- 🔒 100% offline mode with local Whisper, so nothing leaves your machine and no API key is needed

- ✨ Optional cleanup/reformulation that kills the "uhh"s and rewrites the tone

- 🆓 Free and MIT-licensed. Your API keys live in the OS keychain, not on someone's server

Fun fact: Sonora was built almost entirely by an AI agent, with me steering and testing on my own machine. Take it as a little demo of what's possible that way. 🤖

It's early and I'd genuinely love your feedback: which engine you'd pick, what's missing, what breaks. Especially curious to hear from fellow Linux folks who keep getting left out of the dictation party. 🙌

Thanks for checking it out!