Speech Enhancement Based on Cyclegan with Noise-informed Training

10/19/2021
by   Wen-Yuan Ting, et al.
0

Speech enhancement (SE) approaches can be classified into supervised and unsupervised categories. For unsupervised SE, a well-known cycle-consistent generative adversarial network (CycleGAN) model, which comprises two generators and two discriminators, has been shown to provide a powerful nonlinear mapping ability and thus achieve a promising noise-suppression capability. However, a low-efficiency training process along with insufficient knowledge between noisy and clean speech may limit the enhancement performance of the CycleGAN SE at runtime. In this study, we propose a novel noise-informed-training CycleGAN approach that incorporates additional inputs into the generators and discriminators to assist the CycleGAN in learning a more accurate transformation of speech signals between the noise and clean domains. The additional input feature serves as an indicator that provides more information during the CycleGAN training stage. Experiment results confirm that the proposed approach can improve the CycleGAN SE model while achieving a better sound quality and fewer signal distortions.

READ FULL TEXT

page 1

page 2

page 3

page 4

page 5

research
10/19/2021

Speech Enhancement-assisted Stargan Voice Conversion in Noisy Environments

Numerous voice conversion (VC) techniques have been proposed for the con...
research
09/05/2021

A Two-stage Complex Network using Cycle-consistent Generative Adversarial Networks for Speech Enhancement

Cycle-consistent generative adversarial networks (CycleGAN) have shown t...
research
10/27/2022

Audio Signal Enhancement with Learning from Positive and Unlabelled Data

Supervised learning is a mainstream approach to audio signal enhancement...
research
11/02/2022

Inference and Denoise: Causal Inference-based Neural Speech Enhancement

This study addresses the speech enhancement (SE) task within the causal ...
research
10/28/2020

Improving Perceptual Quality by Phone-Fortified Perceptual Loss for Speech Enhancement

Speech enhancement (SE) aims to improve speech quality and intelligibili...
research
04/30/2019

Incorporating Symbolic Sequential Modeling for Speech Enhancement

In a noisy environment, a lossy speech signal can be automatically resto...
research
11/08/2018

Speech Enhancement Based on Reducing the Detail Portion of Speech Spectrograms in Modulation Domain via Discrete Wavelet Transform

In this paper, we propose a novel speech enhancement (SE) method by expl...

Please sign up or login with your details

Forgot password? Click here to reset