Now on Windows.

Hold a key, speak, release, your words appear at the cursor. The same engine that runs TypeMic on Linux and macOS, built natively for Windows, in early access. Dictations run through TypeMic’s own Predator engine, processed, never stored.

• Not yet code-signed. Microsoft is validating our signing certificate. Until it completes, Windows cannot name the publisher, so SmartScreen warns once. The signed build replaces this one the day the certificate is issued. Verify this download the way Linux users verify the .deb: the sha256 and the GPG signature against our release key.
• No automatic updates yet. The updater on Windows refuses to install anything it cannot verify, and verification needs that certificate. New versions are announced here; download and run the installer again.
• A small console window. TypeMic runs in a console window on Windows for now; it shows the status line and the sign-in link. Minimize it, do not close it: closing it stops dictation. The tray icon and background service arrive with the signed build.
• Starts with Windows. Opening TypeMic once registers it to start at sign-in. Uninstall from Settings, Apps, and it stops and removes that entry.

The Windows build is cloud-first: dictation runs through our own Predator engine on Cloudflare’s global infrastructure, which is why the download is ~3 MB rather than hundreds. Audio is processed and immediately discarded, never stored, no voice logs. Windows asks for no permission to watch a key or type at the cursor: running as you is the grant.

Also on Linux, macOS and, in early access, Android. iPhone is next, get notified.