A common question is "What type of permissions the browser will ask for?". When you install and configure your voice tracker, your visitors will be asked to allow speech recognition. It depends if you use HTTPS or HTTP version, this behavior can be slightly different, but in general HTTPS version will make sure that their choice is saved for the next sessions.
Browsers' speech recognition behaves differently based on the protocol used:
https:// - Asks for permission once and remembers the choice.
http:// - Asks for permission repeatedly on every page hit. Results are also returned significantly slower in HTTP.
For excellent user experience, don't compromise on anything less than HTTPS (available free with Let's Encrypt).