- generate dataset with dual faces.
$ docker-compose run dataset python3 2f_2sclean.py -n [number]if you want use another dataset (e.g. audio only, single face, combination of audioset), see other script.
- 0f_1sclean_and_noise.py - no face (audio only), 1 clean speacker + noise from audioset
- 1f_1sclean_and_noise.py - 1 face, 1 clean speacker + noise from audioset
- 2f_2sclean_and_noise.py - 2 face, 2 clean speacker + noise from audioset
- 2s_2sclean.py - 2 face, 2 clean speacker