Separate Passive/Active STT Engines

[GetGoodKid]

I haven't found much documentation about having separate STT engines for passive/active mode.

It seems ideal to have a simple/offline STT (e.g. Sphinx) passively listen for a programmable wake-up word (e.g. "Jasper"). Then it would active listen using a different STT such as Google Speech API. There could be separate passive/active fields in the profile.yml file.

Correct me if I'm wrong, but it seems Jasper currently uses the same STT for both passive and active listening?