Is possible to support with ydotool / nerd-dictation in whisper-ctranslate2 ?

[grigio]

I tried

whisper-ctranslate2 --live_transcribe True  --model small

and live mode works fine, but i'd like to get the output in a non-interactive way, is possible to receive the output of the spoken text after N seconds of silence ?

[jordimas]

You can play with these variables:

https://github.com/Softcatala/whisper-ctranslate2/blob/main/src/whisper_ctranslate2/live.py#L10

These are not available in the cmd, then you need to change the code locally.