I've finally taken the next step with figuring out how to have some sort of decent voice support in Home Assistant. I mean, other than relying on Google Devices.
The picture below shows my completely ridiculous setup for voice recognition:
Total cost: $0, but that's because I reused all the hardware.
I got a RPI4 connected to an iLuv something or other Bluetooth speaker whose battery is completely shot, and a Blue Snowball USB microphone.
I'm running this on the RPI:
https://github.com/rhasspy/wyoming-satellite
Janky as hell, but it works. As I see it, this is really a minimum setup for experimentation. The Atom M5 Echo does not hold a candle.
This is only the beginning.