This repository aims to centralize scripts that prepare datasets to be used with pyannote-audio (more precisely, with its pyannote-database dependency).
Currently available :
To setup each dataset, refer to the README.md contained in their respective folder.
Each dataset comes with its predefined database.yml, containing pyannote-database protocol(s) with already defined train+dev+test sets for out-of-the-box speaker diarization usage.
How these subsets are defined is entirely configurable.
Head to the generate_uris.py of the desired dataset, and edit your_subset_creation_logic().
In particular check compute_uri_subsets_files(...) and compute_uri_subsets_time(...) in scripts/uri.py, which allow you to split according to the number of files or time desired in the subsets.
This split can be absolute (= I want X files in subset1 / I want X hours in subset1) or relative (I want X% of the files in subset1 / I want X% of the hours in subset1).
Don't forget to update the database.yml file accordingly.