Custom Text-to-Speech Voices on Windows 10

The built-in Windows voices aren’t designed to be replaced by simply swapping files.

Windows registers TTS voices through the Speech API (SAPI), so each voice has to be properly installed and registered. The voice registrations are typically stored under:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech\Voices\Tokens

Newer Windows versions may also use:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Speech_OneCore\Voices

The actual voice data is stored in protected system folders, but Microsoft doesn’t support replacing those files directly. Editing or replacing them manually can break TTS functionality or be overwritten by Windows updates. 

If your goal is to use a different voice, the supported approach is to install additional voice packs from Settings > Time & Language > Speech > Manage voices > Add voices. Third-party SAPI-compatible voices can also be installed if they provide a proper installer, and they’ll appear alongside the built-in voices instead of replacing them. 

Out of curiosity, are you trying to install another language, a more natural-sounding voice, or a completely custom voice? That changes what options are available.