The missing voice input for Linux. Beautiful Private Offline
A modern voice input tool built exclusively for Linux. Featuring a transparent "capsule" UI, 100% offline inference (Sherpa-onnx), and native Fcitx5 integration. Sub-20ms latency.
Hi PH community! 👋 I'm Decker, the maker of NexTalk.
For a long time, as a Linux user, I felt a bit... forgotten. While my friends on macOS and Windows enjoyed sleek, integrated voice dictation, our options on Linux were mostly a collection of messy scripts, outdated UIs, or clunky clipboard simulators that broke half the time.
I kept asking myself: "Is a beautiful, high-quality voice experience really a privilege exclusive to Mac and Windows?"
I decided the answer was "No." So I built NexTalk.
My goal was simple: Create a voice input tool that feels like a natural extension of the Linux desktop—Invisible but Powerful.
What makes NexTalk different:
🔒 100% Offline & Private: Your voice never leaves your machine. We use Sherpa-onnx for local inference because your privacy isn't a trade-off for convenience.
🎨 Aesthetics First: A minimalist, transparent capsule UI built with Flutter. No borders, no clutter—just a breathing light that listens.
⚡ Lightning Fast: Sub-20ms latency. It feels like the words are flowing directly from your mind to the screen.
🐧 Linux-Native: No more hacks. We integrated directly with Fcitx5 via Unix Sockets, ensuring perfect compatibility with Wayland and your favorite apps.
NexTalk is my love letter to the Linux community. It's open-source and built for those who refuse to settle for "good enough."
I’d love to hear your thoughts, bug reports, or feature ideas. Let's make Linux productivity better, together! 🚀
