You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
 
 
Morten Kolbæk 142007dc90 Update 'README.md' 2 years ago
utils Update 'utils/losses.py' 2 years ago
MWE_eval.zip Added minimal working example for evaluation 2 years ago
MWE_train.zip Added minimal working example for training 2 years ago
README.md Update 'README.md' 2 years ago
run_model.py Update 'run_model.py' 2 years ago

README.md

On Loss Functions for Supervised Monaural Time-Domain Speech Enhancement

This repository contains the architecture and loss functions studied in [1].

Audio samples are available via https://vbn.aau.dk/files/319712905/demo.zip

Published version is available at: https://ieeexplore.ieee.org/document/8966946

Preprint is available at: https://arxiv.org/abs/1909.01019

If you find this code useful, please consider citing:

@article{kolbaek_loss_2020,
title = {On {Loss} {Functions} for {Supervised} {Monaural} {Time}-{Domain} {Speech} {Enhancement}},
volume = {28},
issn = {2329-9304},
doi = {10.1109/TASLP.2020.2968738},
number = {1},
journal = {IEEE/ACM Transactions on Audio, Speech, and Language Processing},
author = {Kolb{\ae}k, Morten and Tan, Zheng-Hua and Jensen, S{\o}ren Holdt and Jensen, Jesper},
year = {2020},
pages = {825--838}}

The code has been tested using:

To run minimal working example for training: python run_model.py -c ./exp0001.yaml

To run minimal working example for evaluation: matlab eval_predictions.m

If you have any questions or issues regarding the code or paper, don’t hesitate to contact me on LinkedIn (https://www.linkedin.com/in/morten-kolb%C3%A6k-36120b25)

[1] M. Kolbæk, Z. Tan, S. H. Jensen and J. Jensen, "On Loss Functions for Supervised Monaural Time-Domain Speech Enhancement," in IEEE/ACM Transactions on Audio, Speech, and Language Processing, vol. 28, pp. 825-838, 2020.