On a matching GNU/Linux x86-64 or macOS host, follow
Install Omnivox From A Release Binary. That procedure downloads and
verifies the published 1.4.1 archive, keeps its runtime data beside the
executable, and adds its versioned installation directory to PATH. Then run:
omnivox --check
Use Installing Omnivox From Source instead on Linux ARM64, an incompatible Linux distribution, or a host whose security policy does not permit the unsigned macOS archive. No proprietary speech runtime is needed for either GNU/Linux or macOS route.
Under WSL2, follow Install An Omnivox Windows Release Under WSL2 to install and select the verified Windows x64 archive. Then run:
cd /path/where/you/keep/source/emacsvox ./servers/omnivox --check
The check plays a tone followed by speech. When you hear both, the speech server is ready. If either is missing, use Backend Diagnostics And Responsibilities before continuing. See Speech Backends contains the complete platform matrix, build details, and compatibility alternatives. Native Windows Emacs support remains a preview. Evaluators can follow Trying Native Windows Emacsvox; initial users should choose the WSL2 route.