Hold Space anywhere in macOS or Windows 10/11, speak, release — text lands at your cursor. Native Tauri + Rust build, no installer bloat, no account.
DMG · x64 · 16.6 MB · macOS 12+ · Apple Silicon via Rosetta 2
MSI · x64 · 12.5 MB · Windows 10 / 11 · recommended
EXE · x64 · 8.8 MB · NSIS installer · v0.3.0 · 2026-09-22
Local speech recognition by nvidia/parakeet-tdt-0.6b-v3 (CC-BY-4.0) · ~465 MB model, one-click download in Settings.
The demo is not digitally signed yet. macOS: right-click the app → "Open" (Открыть), or run xattr -d com.apple.quarantine /Applications/VoiceDo.app. Windows: SmartScreen → "More info" (Подробнее) → "Run anyway" (Выполнить в любом случае).
On first start, open Settings and add an OpenAI-compatible, Qwen/DashScope or Google key — or skip keys entirely: offline mode runs a local model (~465 MB, one-click download). Google mode works without a key for a quick try.
Default hotkey is configurable in Settings. Hold it in any app — Word, browser, terminal — speak, release, text is typed at the cursor.
The macOS DMG is an x64 build — native on Intel, via Rosetta 2 on Apple Silicon. Builds are published on GitHub Releases with SHA-256 checksums; Apple Developer signing + notarization and Windows code signing are the next step. Meanwhile, source is open: clone and build.