search

wq2012 / SpectralCluster

Python re-implementation of the (constrained) spectral clustering algorithms used in Google's speaker diarization papers.
https://google.github.io/speaker-id/publications/LstmDiarization/
Apache License 2.0
513 stars 73 forks source link

How to not use any single_cluster_condition? #41

Closed hbredin closed 2 years ago

hbredin commented 2 years ago

I am having trouble getting any good results with existing built-in single_cluster_condition.

Therefore, I'd like to try and deactivate it completely. Is there an option for that?

https://github.com/wq2012/SpectralCluster/blob/b6e11827e30e83eedba3f7f659ee634a25095eb6/spectralcluster/fallback_clusterer.py#L55

wq2012 commented 2 years ago

Just set min_clusters != 1?

wq2012 commented 2 years ago

Or use single_cluster_condition=AllAffinity, and set single_cluster_affinity_threshold to a big number?

hbredin commented 2 years ago

Thanks.