SpeechBrain Diarization with Clustering Speakers #2880

Good day! I'm trying to do diarization without any training or inputting speech examples of speakers. So my goal is to do 'anonymous' diarization, where the code defines on his own how many speakers are there in the audio, and which segment to whom belongs (Speaker 1, Speaker 2, Speaker 3, etc.)
In my research I've found module speechbrain.processing.diarization, which is answering my needs. However, I couldn't use it, since Classes return... nothing. Nothing or 'self', but they do not have any attributes where they store the result. So far I've tried two classes Spec_Cluster and Spec_Clust_unorm. But their clustering functions don't return results, so I am quite confused. Is it a bug? Or I am using them completely wrong?
I know that there're some other clustering functions, but I'm quite confused about the input data they need.

diary_obj (StatObject_SB type) – Contains embeddings in diary_obj.stat1 and segment IDs in diary_obj.segset.