agile_squirrel

I haven’t verified it, but it local ML models on audio is pretty efficient and should run fine on a raspberry pi.