whspr v0.1.0

whspr — hold a hotkey, speak, and clean text is injected into whatever app has focus. Local Whisper transcription + optional local-LLM cleanup, in a Modernist desktop UI.

First release, on both desktop OSes and both CPU architectures.

Windows

File	For
whspr-0.1.0-windows-x64-setup.exe	Windows 10/11 · x64 (Intel/AMD)	Installer — double-click. Per-user install into %LOCALAPPDATA%\whspr, Start-menu shortcut, optional launch-at-sign-in. No admin / no UAC.
whspr-0.1.0-windows-arm64-setup.exe	Windows 10/11 · ARM (Snapdragon etc.)	Installer — native aarch64. Same per-user install, self-contained.

Windows ships as an installer only (the installer provisions the per-user app folder). Speaker diarization ("Speakers") is unavailable on arm64 (no sherpa-onnx prebuilt for that target); everything else — dictation, local-LLM cleanup, History, Models — works. The build is unsigned, so SmartScreen may warn on first launch (More info → Run anyway).

macOS

File	For
whspr-0.1.0-macos-universal.dmg	Apple Silicon (M1+) and Intel — one universal build, native on both

Drag whspr.app to Applications. Unsigned, so on first launch right-click → Open (or System Settings → Privacy & Security → Open Anyway). On first run it asks for microphone + accessibility permission.

Using it

• Push-to-talk: Ctrl+Shift+Space on Windows, Ctrl+Space on macOS. Hold, speak, release — the cleaned text types itself into whatever app has focus.
• Speech/LLM models download on first use from the Models screen. whspr lives in the system tray.