i run HA through docker, and from what i can tell, you would need to run different piper instances in order to load different voices, then add them in to the wyoming integration. i’m not sure of that, but seems like a workable option. i might give this a try tonight or tomorow. if both instance responded at the same time, it might be a much for the processor, depending on what you host on.

that seems like its exactly what i’m looking for, guess i know what my project for the weekend is gonna be. thanks! 😀

looks pretty interesting, but it’s based on invidious, i’ll have to keep an eye out to see how this works out. https://news.slashdot.org/story/23/06/15/1956200/youtube-tells-open-source-privacy-software-invidious-to-shut-down