vosk-android-demo VS spokestack-android

If Dicio glitches for you (it has trouble downloading the models sometimes), you can try the Vosk demo app instead. It will just print out what you say.

No, however there is the Vosk Android Demo Keyboard It's more of a proof of concept than it is practical, but its functional if you need that feature.

There is the idea to wrap the lbrary in an "input device" for android but nothing concrete

Hey, contrary to this, I've found an open source android voice recognition software that works offline, it seems to be very lightweight, it's called Vosk https://github.com/alphacep/vosk-android-demo

Spokestack is on a mission to bring private, personalized voice tech to mobile apps, and we're beta testing our newest AutoML service. I can't divulge all its features publicly yet, but they add a whole new dimension to what you can already do in an app using our Android library and tutorials.

Good stuff. Nice to see Tasker and Termux being considered as clients upfront. Maybe you'd also get some use out of Spokestack's integrations: we* have a modular system for hooking into different services for different parts of the request pipeline (wakeword/ASR/NLU/TTS). Android ASR and Spokestack NLU work offline like your assistant framework; the rest would require an Internet connect.