Stationary/transient audio separation using convolutional autoencoders

Gerard Roma; Owen Green; Pierre Alexandre Tremblay

Stationary/transient audio separation using convolutional autoencoders

Gerard Roma, Owen Green, Pierre Alexandre Tremblay

Research Centre for Performance Practices

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Extraction of stationary and transient components from audio has many potential applications to audio effects for audio content pro- duction. In this paper we explore stationary/transient separation using convolutional autoencoders. We propose two novel unsuper- vised algorithms for individual and and joint separation. We de- scribe our implementation and show examples. Our results show promise for the use of convolutional autoencoders in the extraction of sparse components from audio spectrograms, particularly using monophonic sounds.

Original language	English
Title of host publication	Proceedings of the 21st International Conference on Digital Audio Effects (DAFx-18)
Pages	65-71
Number of pages	7
Publication status	Published - 4 Sep 2018
Event	21st International Conference on Digital Audio Effects - Universidade de Aveiro, Aveiro, Portugal Duration: 4 Sep 2018 → 8 Sep 2018 http://dafx2018.web.ua.pt/ (Link to Conference Website)

Publication series

Name
ISSN (Print)	2413-6700
ISSN (Electronic)	2413-6689

Conference

Conference	21st International Conference on Digital Audio Effects
Abbreviated title	DAFx-18
Country	Portugal
City	Aveiro
Period	4/09/18 → 8/09/18
Internet address	http://dafx2018.web.ua.pt/ (Link to Conference Website)

Access to Document

DAFx2018_paper_61Final published version, 8.1 MBLicence: Unspecified

http://dafx2018.web.ua.pt/papers/DAFx2018_paper_61.pdfLicence: Unspecified

1 Conference contribution

Interdisciplinary Research as Musical Experimentation: A case study in musicianly approaches to sound corpora
Green, O., Tremblay, P. A. & Roma, G., 24 Jan 2019, Proceedings of the Electroacoustic Music Studies Network Conference . Zenodo, 12 p.
Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Open Access

1 Active

FluCoMa: Fluid Corpus Manipulations
Tremblay, P. A., Green, O., Roma, G., Harker, A., Clarke, M. & Dufeu, F.
1/09/17 → 31/08/22
Project: Research

Cite this

@inproceedings{4f6a24f95f934f85a3fa78257b3045b8,

title = "Stationary/transient audio separation using convolutional autoencoders",

abstract = "Extraction of stationary and transient components from audio has many potential applications to audio effects for audio content pro- duction. In this paper we explore stationary/transient separation using convolutional autoencoders. We propose two novel unsuper- vised algorithms for individual and and joint separation. We de- scribe our implementation and show examples. Our results show promise for the use of convolutional autoencoders in the extraction of sparse components from audio spectrograms, particularly using monophonic sounds.",

author = "Gerard Roma and Owen Green and Tremblay, {Pierre Alexandre}",

year = "2018",

month = sep,

day = "4",

language = "English",

pages = "65--71",

booktitle = "Proceedings of the 21st International Conference on Digital Audio Effects (DAFx-18)",

note = "21st International Conference on Digital Audio Effects, DAFx-18 ; Conference date: 04-09-2018 Through 08-09-2018",

url = "http://dafx2018.web.ua.pt/",

}

TY - GEN

T1 - Stationary/transient audio separation using convolutional autoencoders

AU - Roma, Gerard

AU - Green, Owen

AU - Tremblay, Pierre Alexandre

PY - 2018/9/4

Y1 - 2018/9/4

N2 - Extraction of stationary and transient components from audio has many potential applications to audio effects for audio content pro- duction. In this paper we explore stationary/transient separation using convolutional autoencoders. We propose two novel unsuper- vised algorithms for individual and and joint separation. We de- scribe our implementation and show examples. Our results show promise for the use of convolutional autoencoders in the extraction of sparse components from audio spectrograms, particularly using monophonic sounds.

AB - Extraction of stationary and transient components from audio has many potential applications to audio effects for audio content pro- duction. In this paper we explore stationary/transient separation using convolutional autoencoders. We propose two novel unsuper- vised algorithms for individual and and joint separation. We de- scribe our implementation and show examples. Our results show promise for the use of convolutional autoencoders in the extraction of sparse components from audio spectrograms, particularly using monophonic sounds.

M3 - Conference contribution

SP - 65

EP - 71

BT - Proceedings of the 21st International Conference on Digital Audio Effects (DAFx-18)

T2 - 21st International Conference on Digital Audio Effects

Y2 - 4 September 2018 through 8 September 2018

ER -

Stationary/transient audio separation using convolutional autoencoders

Abstract

Publication series

Conference

Access to Document

Fingerprint

Research output

Interdisciplinary Research as Musical Experimentation: A case study in musicianly approaches to sound corpora

Projects

FluCoMa: Fluid Corpus Manipulations

Cite this