Speaker-segmentation pipeline source code

[Spectra456]

Hi, I'm trying to obtain Voice Activity Detection (VAD) results combined with Speaker Change Detection. I found that the pyannote/speaker-segmentation pipeline seems to be the best fit for my purposes. However, I don't fully understand what exactly happens inside this pipeline, as I couldn't find the relevant source code in the PyAnnote repository. Could you help me understand it better?

I tried replicating the results based on this tutorial https://herve.niderb.fr/fastpages/2022/10/23/One-speaker-segmentation-model-to-rule-them-all.html, but the outcomes were very different, even with the same hyperparameters. Thanks a lot for your help!