Generative adversarial network-based approach to signal reconstruction from magnitude spectrogram

Keisuke Oyamada, Hirokazu Kameoka, Takuhiro Kaneko, Kou Tanaka, Nobukatsu Hojo, Hiroyasu Ando

Research output: Chapter in Book/Report/Conference proceedingConference contribution

12 Citations (Scopus)

Abstract

In this paper, we address the problem of reconstructing a time-domain signal (or a phase spectrogram) solely from a magnitude spectrogram. Since magnitude spectrograms do not contain phase information, we must restore or infer phase information to reconstruct a time-domain signal. One widely used approach for dealing with the signal reconstruction problem was proposed by Griffin and Lim. This method usually requires many iterations for the signal reconstruction process and depending on the inputs, it does not always produce high-quality audio signals. To overcome these shortcomings, we apply a learning-based approach to the signal reconstruction problem by modeling the signal reconstruction process using a deep neural network and training it using the idea of a generative adversarial network. Experimental evaluations revealed that our method was able to reconstruct signals faster with higher quality than the Griffin-Lim method.

Original languageEnglish
Title of host publication2018 26th European Signal Processing Conference, EUSIPCO 2018
PublisherEuropean Signal Processing Conference, EUSIPCO
Pages2514-2518
Number of pages5
ISBN (Electronic)9789082797015
DOIs
Publication statusPublished - 2018 Nov 29
Externally publishedYes
Event26th European Signal Processing Conference, EUSIPCO 2018 - Rome, Italy
Duration: 2018 Sep 32018 Sep 7

Publication series

NameEuropean Signal Processing Conference
Volume2018-September
ISSN (Print)2219-5491

Conference

Conference26th European Signal Processing Conference, EUSIPCO 2018
Country/TerritoryItaly
CityRome
Period18/9/318/9/7

Keywords

  • Deep neural networks
  • Generative adversarial networks
  • Phase reconstruction

ASJC Scopus subject areas

  • Signal Processing
  • Electrical and Electronic Engineering

Fingerprint

Dive into the research topics of 'Generative adversarial network-based approach to signal reconstruction from magnitude spectrogram'. Together they form a unique fingerprint.

Cite this