Speech Antispoofing Project

Based on this template

Installation guide

pip install -r ./requirements.txt
mkdir default_test_model
wget https://datashare.ed.ac.uk/bitstream/handle/10283/3336/LA.zip
unzip LA.zip

Usage guide

To train

python3 train.py --config-name lcnn_cqt ++kaiman=False ++loss.weight=[] ++lr_scheduler.gamma=0.95 ++seed=0

To test:

gdown https://drive.google.com/file/d/1fBgiJCJ9jO1Ab_J0WKf2097-x8fqRREb/view?usp=sharing -O default_test_model/checkpoint.pth --fuzzy
python3 test.py

The data will be logged to wandb. You can change wavs in `default_test_data`` if you want.