Pytorch model for voice filtering trained with LibriSpeech
1)Embedder chkpt: https://drive.google.com/file/d/1LCdZNsvx6BEzVzy50BSCRAtvbrZtE6sV/view?usp=sharing
2)Filter chkpt(mse loss, train 1000h on LS 61k epochs, 8 days with gtx 1080ti) Google Drive: https://drive.google.com/file/d/1RRDGBFnyL5po1VXpkfJ_yRQryZ6fEWPW/view?usp=sharing
Examples for this checkpoint are in lib "Dialogs test"
After downloading both chkpts you can try system:
python3 inf.py -c [config yaml] -e [path of embedder pt file] --checkpoint_path [path of chkpt pt file] -m [path of mixed wav file] -r [path of reference wav file] -o [output directory]
Train history:
