A Brief Overview of Unsupervised Neural Speech Representation Learning

Lasse Borgholt; Jakob Drachmann Havtorn; Joakim Edin; Lars Maaløe; Christian Igel

A Brief Overview of Unsupervised Neural Speech Representation Learning

Lasse Borgholt
, Jakob Drachmann Havtorn
, Joakim Edin
, Lars Maaløe
, Christian Igel

Corti
University of Copenhagen

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

229 Downloads (Orbit)

Abstract

Unsupervised representation learning for speech processing has matured greatly in the last few years. Work in computer vision and natural language processing has paved the way, but speech data offers unique challenges. As a result, methods from other domains rarely translate directly. We review the development of unsupervised representation learning for speech over the last decade. We identify two primary model categories: self-supervised methods and probabilistic latent variable models. We describe the models and develop a comprehensive taxonomy. Finally, we discuss and compare models from the two categories.

Original language	English
Title of host publication	Proceedings of 2^nd Workshop on Self-supervised Learning for Audio and Speech Processing
Number of pages	13
Publisher	Association for the Advancement of Artificial Intelligence
Publication date	2022
Publication status	Published - 2022
Event	2^nd Workshop on Self-supervised Learning for Audio and Speech Processing - Virtual Venue Duration: 22 Feb 2022 → 1 Mar 2022 Conference number: 2 https://aaai-sas-2022.github.io/

Conference

Conference	2^nd Workshop on Self-supervised Learning for Audio and Speech Processing
Number	2
Location	Virtual Venue
Period	22/02/2022 → 01/03/2022
Internet address	https://aaai-sas-2022.github.io/

Access to Document

FulltextSubmitted manuscript, 963 KB

OpenUrl availability

Check availability in DTU Findit

Cite this

@inproceedings{dde1282968c64712804d0e7943d830a6,

title = "A Brief Overview of Unsupervised Neural Speech Representation Learning",

abstract = "Unsupervised representation learning for speech processing has matured greatly in the last few years. Work in computer vision and natural language processing has paved the way, but speech data offers unique challenges. As a result, methods from other domains rarely translate directly. We review the development of unsupervised representation learning for speech over the last decade. We identify two primary model categories: self-supervised methods and probabilistic latent variable models. We describe the models and develop a comprehensive taxonomy. Finally, we discuss and compare models from the two categories.",

author = "Lasse Borgholt and Havtorn, \{Jakob Drachmann\} and Joakim Edin and Lars Maal{\o}e and Christian Igel",

year = "2022",

language = "English",

booktitle = "Proceedings of 2nd Workshop on Self-supervised Learning for Audio and Speech Processing",

publisher = "Association for the Advancement of Artificial Intelligence",

note = "2<sup>nd</sup> Workshop on Self-supervised Learning for Audio and Speech Processing, AAAI SAS 2022 ; Conference date: 22-02-2022 Through 01-03-2022",

url = "https://aaai-sas-2022.github.io/",

}

Borgholt, L, Havtorn, JD, Edin, J, Maaløe, L & Igel, C 2022, A Brief Overview of Unsupervised Neural Speech Representation Learning. in Proceedings of 2^nd Workshop on Self-supervised Learning for Audio and Speech Processing. Association for the Advancement of Artificial Intelligence, 2^nd Workshop on Self-supervised Learning for Audio and Speech Processing, 22/02/2022.

A Brief Overview of Unsupervised Neural Speech Representation Learning. / Borgholt, Lasse; Havtorn, Jakob Drachmann; Edin, Joakim et al.
Proceedings of 2^nd Workshop on Self-supervised Learning for Audio and Speech Processing. Association for the Advancement of Artificial Intelligence, 2022.

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

TY - GEN

T1 - A Brief Overview of Unsupervised Neural Speech Representation Learning

AU - Borgholt, Lasse

AU - Havtorn, Jakob Drachmann

AU - Edin, Joakim

AU - Maaløe, Lars

AU - Igel, Christian

N1 - Conference code: 2

PY - 2022

Y1 - 2022

N2 - Unsupervised representation learning for speech processing has matured greatly in the last few years. Work in computer vision and natural language processing has paved the way, but speech data offers unique challenges. As a result, methods from other domains rarely translate directly. We review the development of unsupervised representation learning for speech over the last decade. We identify two primary model categories: self-supervised methods and probabilistic latent variable models. We describe the models and develop a comprehensive taxonomy. Finally, we discuss and compare models from the two categories.

AB - Unsupervised representation learning for speech processing has matured greatly in the last few years. Work in computer vision and natural language processing has paved the way, but speech data offers unique challenges. As a result, methods from other domains rarely translate directly. We review the development of unsupervised representation learning for speech over the last decade. We identify two primary model categories: self-supervised methods and probabilistic latent variable models. We describe the models and develop a comprehensive taxonomy. Finally, we discuss and compare models from the two categories.

M3 - Article in proceedings

BT - Proceedings of 2nd Workshop on Self-supervised Learning for Audio and Speech Processing

PB - Association for the Advancement of Artificial Intelligence

T2 - 2<sup>nd</sup> Workshop on Self-supervised Learning for Audio and Speech Processing

Y2 - 22 February 2022 through 1 March 2022

ER -

A Brief Overview of Unsupervised Neural Speech Representation Learning

Abstract

Conference

Access to Document

OpenUrl availability

Fingerprint

Cite this