RawNet2 implementation

Installation guide

1. clone repository

git clone https://github.com/wh1tePigeon/RawNet2

2. install requirements

pip install -r ./requirements.txt

Train

Download dataset and specify path to train, dev and eval files and protocols in config. Then run

python3 train.py -c *path_to_config*

Main model was trained with source/configs/config_kaggle.json.

Download checkpoints

Checkpoints and training logs can be found here. To download them use download_model.py script. If you want to test models other than main, make sure you edited code according to model.

Test

To test solution run

python3 test.py

Wandb

Wandb report

Credits

This repository is based on a heavily modified fork of hw template repository.