Seems a really interesting speech framework with a specific ASR type application here
New to me but seems to be getting a lot of traction especially for embedded / micro applications
I noticed it on the mlcommons github but still waxing lyrical about their spoken word datasets prob some people on here might want to get involved.