Looking at options for cloud-free on-device voice control and getting quite dispirited by the general state of things. Strong impression is that it's all binary blobs and cloudbased services in the build process held together with lashings of python

self reply