Hi all! New here, but thought I would share something I just built primarily to use with HA. It’s a wyoming server for the Kokoro TTS model. I was looking for a good Jarvis impersonation, and Kokoro’s em_lewis is pretty good, at least to my ears, so I wrapped it in a wyoming server. Works pretty well for me with HA, but that’s pretty much the extent of the testing. 
Take a look if you want to try it out. GitHub - dstrout/wyoming-kokoro: A wyomin test-to-speech (TTS) server using the kokoro TTS model and voices