First Windows alpha. v0.1.0-alpha.1 is a developer-only prerelease. This is not a public release, and vocawin.com stays Coming soon.
The app turns speech into text on your own PC. Audio and models stay here. Nothing is sent to a cloud transcription service.
Installers
Unsigned NSIS and MSI are attached to this prerelease.
Windows will likely say the publisher is unknown. That is expected until we sign. There is no public download page and no auto-update.
Using it
Hold Right Alt to dictate, or switch to toggle in Settings. The shortcut applies as soon as you save. AltGr still types the character your layout expects.
Open Models to download an engine. Whisper Tiny is selected on a new install when it is already on disk. Whisper can use Vulkan if the SDK is present. ONNX models (Parakeet, SenseVoice, Canary, GigaAM, and the rest) unpack from their archives. Moonshine Tiny installs as loose files.
The tray icon is the same white mic. The square behind it is teal when idle, red while the mic is hot, and amber while it transcribes.
The window title says VocaWin · Alpha. The sidebar has a small Alpha tag.
Fixes that landed here
Saving settings used to fail if launch-at-login was off and Windows had no startup shortcut. Start used to hang the window. ONNX downloads used to delete their own archive. Those are fixed.
This build also has local history, logs, a mic test, and auto-pause when a watched app is running.
If something breaks, open an issue on https://github.com/VocaHQ/vocawin/issues.