A little Google search turned up a possible
$1 solution to your question as the first result. If you use headphones you might be able to route the screen-reader audio through them while letting everything else be capture by the video conferencing client, which isn't exactly what you asked for but in my opinion is a superior option.
Another little Google search turned up yet another solution to
$1 also as the first result where someone suggests creating a profile where you just disable the synthesizer, but in my opinion the first solution is better.
Disclaimer: I do not have access to a Windows machine, and haven't used Windows and NVDA since 2019, therefore I was unable to verify any of this.