VoicePrivacy 2020 for anonymization system traing consists of subsets from the following corpora:LibriSpeech - train-clean-100, train-other-500,LibriTTS - train-clean-100, train-other-500 and VoxCeleb 1 & 2 - all. For development and evaluation it consists VCTK - subsets vctk_dev and vctk_test, LibriSpeech - subsets libri_dev and libri_test.

The VoicePrivacy 2020 Challenge Evaluation Plan