I’ve actually been looking into building a voice assistant setup that: a) operates without the need for any internet whatsoever (entirely on-prem); b) is able to access, say, a local file share or NAS store where my audio collection resides and play it on one or more devices; and c) interact with my home automation system (a Hubitat).
I’ve got a couple ideas for it, but I need to acquire/build the hardware for it, which is sadly not going to be cheap.