DeePMOS

Authors: Xinyu Liang, Fredrik Cumlin
Emails: hopeliang990504@gmail.com, fcumlin@gmail.com

Data Preparation

For VCC2018 data, it can be downloaded from here.
For BVCC data, here's the link.

Training

Run train.py, which depend on the following arguments:

• --num_epochs: The number of epochs during training.
• --lamb_c: Weight of consistency loss lambda_c.
• --lamb_t: Weight of teacher model loss lambda_t.
• --log_valid: The number of epochs between logging results on validation data.
• --log_epoch: The number of epochs between logging training losses.
• --dataset: Dataset. 'vcc2018' or 'bvcc'.
• --data_path: Path to the dataset folder.
• --id_table: Path to the id_table folder.
• --save_path: Path for the best model to be saved.

Acknowledgement

This repository inherits from the unofficial MBNet implementation, the LDNet implementation and LaMOSNet implementation.