v2.0

• Speaker ids with respective genders added.
• The speakers of train, test and dev corpora do not overlap.
• A major increase in size with a total time of 611 hours 43 minutes.
• File structure resembles Common Voice data for compatibility.

The version 2.0 corpus can be downloaded in parts:

• clean_train
• clean_test
• clean_dev
• other_train
• other_test
• other_dev