When forward “example.wav”, Can not get the same result as Readme about datadriven-gpvad HOT 4 CLOSED

Hey thanks,
it's actually due to the commit today #9 .
As you already noted, the offset is wrong, which simply means that the sampling rates are not correct.

I just fixed that, it was just the indent of a single line in forward.py.

I did a fresh pull and ran the code and got the following:

Thanks for noticing!

from datadriven-gpvad.

Hey there, probably not, I forgot which model I used for that README. Also I fixed a bug that was there before, which used a wrong sampling rate, so that might be it. Even though that sampling rate error is generally not a big problem.

from datadriven-gpvad.

Thanks. By the way, is "onset" and "offset" means second? If that, the VAD result could be shown bellow, which means drop a lot of speech.

Also, I tried all pretrained models, the biggest offset is 1.92, which also means drop some speech.

from datadriven-gpvad.

Thank you very much!

from datadriven-gpvad.