Combined profiles to avoid having to switch between languages

[yorrd]

i'm using voice2json to implement my Own version of the mac speech the text feature for linux. Since i am a germans speaker, I regularly switch Between english and German.

Is it a good idea to just Run both profiles (de and en) and then compare the prediction scores? do you think this should be implemented within this repository or by just running two docker containers, One for each profile? if you think this belongs here, i'm happy to work on a pull request.

---

I appreciate the work a Lot. Using this repository to dictate this Issue as well. I wrote a Small Linux Front End that basically uses the clipboard to find out its surroundings and inputs dictated text sequences accordingly. As you can see, I added a truecaser which is trained on a German dataset, so it's producing crappy results.