by the way, I'm planning to switch two speech to text for most of my typing needs in the near future so expect my grammar to have a bunch of misplaced words.
also, I'm gonna stop bothering with capitalizing for the first letter of each sentence since I mostly using this for programming and that's adding the code for it would make coding more annoying.
@mauve I'm probably diving into a speech driven project later this year. What recognition engine are you using and do you like it?
@trevorflowers I'm using VOSK wrapped by https://github.com/ideasman42/nerd-dictation
It isn't amazing but it's fast enough and offline first. The python scripting also makes it easy to hack on things.
I'm publishing my scripts for it here: https://github.com/RangerMauve/mauve-dictation/blob/default/nerd-dictation.py (mostly basic stuff like special characters for programming and some aliases for common mistypings).
I find that I do a moderate amount of manual correction where I select messed up words and speak the word again to overwrite them.
@mauve Related: I just found out about Rhasspy which looks like more of a VUI foundation than a text input system but it interfaces with a ton of useful voice libs.
https://github.com/rhasspy/rhasspy3/
@trevorflowers Oh yeah, I saw that before but ended up not getting into it because it had too many moving pieces. 😅 looks pretty solid thougb
@mauve Thank you for the pointers. I'll give them a look.