The dataset this project is based on is VoxCeleb.

We have written a script to assist in downloading the data. The script requires:

• Python 3.5+
• tqdm==4.17.1
• youtube-dl
• ffmpeg

To use the script, you will first need to download the VoxCeleb dataset. Then run:

./download.py /location/to/voxceleb1_txt /location/to/output_dir

Any download errors will be logged in a generated file.