If this works, what is the process?
#Is it possible to use Open WakeWord models streaming to a Voice Preview Edition?
charred elbow
wind flicker
If this works, what is the process?
so... technically yes it is possible but you would have to modify the firmware. I don't think the procedure is documented anywhere directly but its not doable. there have been a couple of custom firmware configs for the VPE but they never seem to keep up to date. so you end up loosing other functionality or not getting new functionality. so if your gunna go down that road then learning how to do it is the better option.
realistically, if you are looking for a simple method that doesn't require some work then its not really doable.
gilded dew
Yeah as Michael said - it's possible. But the device is loosing a lot of functionality (like "stop" word or proper LED stages signalling), so not that many tinkerers there.
You probably better off with custom word for MWW - I saw GitHub repos for that.
bitter aurora
https://github.com/TaterTotterson/microWakeWord-Trainer-Nvidia-Docker
This can be used to generate a microwakeword!
It takes about 1 hour on a 3090