Speech input is one of the missing features in #Phosh's stevia. I had looked at several possible solutions but didn't want to pull in a ton more dependencies into stevia itself.
While looking for something completely different I stumbled onto #vosk-server which runs fully locally but can be talked to via websocket and so I could punch that into the prototype I had already alying around (video has audio):
There's more work needed to make this usable (we don't have arm64 docker containers of vosk-server and it's not in any distros but most of it is in alpine already). There's also room for improvement regarding the recognition (it likes to guess "the" when there's no input as you can see at the end of the video).
If you want to help out, jump into https://gitlab.gnome.org/World/Phosh/stevia/-/merge_requests/279 .
Thanks to dogman in the FLX1 channel for mentioning vosk which led me to vosk-server.
Guido Günther
Manuel Traut 🐧