search

ideasman42 / nerd-dictation

Simple, hackable offline speech to text - using the VOSK-API.
GNU General Public License v3.0
1.27k stars 106 forks source link

--output STDOUT only sends output after dictation ends #109

Closed jtara1 closed 9 months ago

jtara1 commented 9 months ago

Is this the intention or a bug?

./nerd-dictation begin \
  --numbers-as-digits \
  --numbers-no-suffix \
  --output STDOUT
jtara1 commented 9 months ago

I noticed it can delete many words to retype them when using the default --output. I'm guessing this is a feature of vosk with support for the retyping through xdotool.

This could then mean that the answer to my question is its the intended way it works as you can't change what's already sent to stdout.

jtara1 commented 9 months ago

The startup time loading an improved bigger model is about 7s so calling begin and end on the dictation process is a little annoying if I just want intermittent results.

jtara1 commented 9 months ago

I just wanted a more programmatic way (i.e.: dictation piped to a socket) to work with the output of the live dictation. Maybe it's worth exploring the use of nerd-dictation in a VM or re-evaluating the output of vosk to find the best way to work with it.