Hell yeah self host that video converting software yourself!

Also neural network upscaling always makes weird little faces on people far away, which is pretty disconcerting for me. Just isn’t good enough

Have you thought about https://github.com/oyvindln/vhs-decode for your capture method? The results from direct signal capture look damn good

It’s iOS-specific, but if your family happen to have iOS devices https://github.com/mikebrady/shairport-sync works amazingly with Rpi ZeroWs. But I’d spring for the 2W version for some breathing room, if you could. All the iOS apps I’ve used are able to play to an airplay2 group in sync, but I don’t have either of the apps you called out specifically.