vosk-api icon indicating copy to clipboard operation
vosk-api copied to clipboard

Feature: setting the required silence length

Open johngebbie opened this issue 1 year ago • 2 comments

It would be great to be able to set the required silence/pause length after a sentence programmatically.

I reduced the min-trailing-silence rules in the model.conf and it feels great typing keys with Numen, but it also affects the literal transcription which uses the same model, and I don't want to load two copies of it (or mess with the models really).

Perhaps the API could be:

rec.SetSilence1(0.123)
rec.SetSilence2(0.123)
rec.SetSilence3(0.123)
rec.SetSilence4(0.123)

for Kaldi's rule1/2/3/4. But just being able to set rule2 would be very nice.

Related issues: https://github.com/alphacep/vosk-api/issues/1329 https://github.com/alphacep/vosk-api/issues/380

johngebbie avatar May 02 '23 17:05 johngebbie

Sure, we can do something like that. Let me look coming days.

nshmyrev avatar May 03 '23 09:05 nshmyrev

+1

gregtzar avatar Feb 17 '24 19:02 gregtzar