This repository automatically downloads the MSDWILD dataset and set it up to be used with pyannote-database.
Clone this repository, download the dataset zip at https://github.com/X-LANCE/MSDWILD#wavs and put it under the msdwild
folder.
Then, run setup.sh
in the msdwild
directory to download/extract/generate the files (wav, rttm, uem, uris).
If you want to edit how custom subsets are generated, head to generate_uris.py where you can edit them through constants at the beginning of the file. If you add/remove subsets, don't forget to edit database.yml accordingly.