Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS)

Ben Foley; Josh Arnold; Rolando Coto-Solano; Gautier Durantin; T. Mark Ellison; Daan van Esch; Scott Heath; František Kratochvíl; Zara Maxwell-Smith; David Nash; Ola Olsson; Mark Richards; Nay San; Hywel Stoakes; Nick Thieberger; Janet Wiles

doi:10.21437/SLTU.2018-43

Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS)

Ben Foley, Josh Arnold, Rolando Coto-Solano, Gautier Durantin, T. Mark Ellison, Daan van Esch, Scott Heath, František Kratochvíl, Zara Maxwell-Smith, David Nash, Ola Olsson, Mark Richards, Nay San, Hywel Stoakes, Nick Thieberger, Janet Wiles

Research output: Chapter in Book / Conference Paper › Conference Paper

43 Citations (Scopus)

Abstract

Machine learning has revolutionized speech technologies for major world languages, but these technologies have generally not been available for the roughly 4,000 languages with populations of fewer than 10,000 speakers. This paper describes the development of ELPIS, a pipeline which language documentation workers with minimal computational experience can use to build their own speech recognition models, resulting in models being built for 16 languages from the Asia-Pacific region. ELPIS puts machine learning speech technologies within reach of people working with languages with scarce data, in a scalable way. This is impactful since it enables language communities to cross the digital divide, and speeds up language documentation. Complete automation of the process is not feasible for languages with small quantities of data and potentially large vocabularies. Hence our goal is not full automation, but rather to make a practical and effective workflow that integrates machine learning technologies.

Original language	English
Title of host publication	Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India
Publisher	International Speech Communication Association
Pages	205-209
Number of pages	5
DOIs	https://doi.org/10.21437/SLTU.2018-43
Publication status	Published - 2018
Event	International Workshop on Spoken Language Technologies for Under-Resourced Languages - Duration: 1 Jan 2018 → …

Conference

Conference	International Workshop on Spoken Language Technologies for Under-Resourced Languages
Period	1/01/18 → …

Keywords

automatic speech recognition
endangered languages

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.21437/SLTU.2018-43

Cite this

Foley, B., Arnold, J., Coto-Solano, R., Durantin, G., Ellison, T. M., van Esch, D., Heath, S., Kratochvíl, F., Maxwell-Smith, Z., Nash, D., Olsson, O., Richards, M., San, N., Stoakes, H., Thieberger, N., & Wiles, J. (2018). Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS). In Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India (pp. 205-209). International Speech Communication Association. https://doi.org/10.21437/SLTU.2018-43

Foley, Ben ; Arnold, Josh ; Coto-Solano, Rolando et al. / Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS). Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India. International Speech Communication Association, 2018. pp. 205-209

@inproceedings{17ef2135228941b4a76a319dca6d8a37,

title = "Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS)",

abstract = "Machine learning has revolutionized speech technologies for major world languages, but these technologies have generally not been available for the roughly 4,000 languages with populations of fewer than 10,000 speakers. This paper describes the development of ELPIS, a pipeline which language documentation workers with minimal computational experience can use to build their own speech recognition models, resulting in models being built for 16 languages from the Asia-Pacific region. ELPIS puts machine learning speech technologies within reach of people working with languages with scarce data, in a scalable way. This is impactful since it enables language communities to cross the digital divide, and speeds up language documentation. Complete automation of the process is not feasible for languages with small quantities of data and potentially large vocabularies. Hence our goal is not full automation, but rather to make a practical and effective workflow that integrates machine learning technologies.",

keywords = "automatic speech recognition, endangered languages",

author = "Ben Foley and Josh Arnold and Rolando Coto-Solano and Gautier Durantin and Ellison, {T. Mark} and {van Esch}, Daan and Scott Heath and Franti{\v s}ek Kratochv{\'i}l and Zara Maxwell-Smith and David Nash and Ola Olsson and Mark Richards and Nay San and Hywel Stoakes and Nick Thieberger and Janet Wiles",

year = "2018",

doi = "10.21437/SLTU.2018-43",

language = "English",

pages = "205--209",

booktitle = "Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India",

publisher = "International Speech Communication Association",

note = "International Workshop on Spoken Language Technologies for Under-Resourced Languages ; Conference date: 01-01-2018",

}

Foley, B, Arnold, J, Coto-Solano, R, Durantin, G, Ellison, TM, van Esch, D, Heath, S, Kratochvíl, F, Maxwell-Smith, Z, Nash, D, Olsson, O, Richards, M, San, N, Stoakes, H, Thieberger, N & Wiles, J 2018, Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS). in Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India. International Speech Communication Association, pp. 205-209, International Workshop on Spoken Language Technologies for Under-Resourced Languages, 1/01/18. https://doi.org/10.21437/SLTU.2018-43

Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS). / Foley, Ben; Arnold, Josh; Coto-Solano, Rolando et al.
Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India. International Speech Communication Association, 2018. p. 205-209.

Research output: Chapter in Book / Conference Paper › Conference Paper

TY - GEN

T1 - Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS)

AU - Foley, Ben

AU - Arnold, Josh

AU - Coto-Solano, Rolando

AU - Durantin, Gautier

AU - Ellison, T. Mark

AU - van Esch, Daan

AU - Heath, Scott

AU - Kratochvíl, František

AU - Maxwell-Smith, Zara

AU - Nash, David

AU - Olsson, Ola

AU - Richards, Mark

AU - San, Nay

AU - Stoakes, Hywel

AU - Thieberger, Nick

AU - Wiles, Janet

PY - 2018

Y1 - 2018

N2 - Machine learning has revolutionized speech technologies for major world languages, but these technologies have generally not been available for the roughly 4,000 languages with populations of fewer than 10,000 speakers. This paper describes the development of ELPIS, a pipeline which language documentation workers with minimal computational experience can use to build their own speech recognition models, resulting in models being built for 16 languages from the Asia-Pacific region. ELPIS puts machine learning speech technologies within reach of people working with languages with scarce data, in a scalable way. This is impactful since it enables language communities to cross the digital divide, and speeds up language documentation. Complete automation of the process is not feasible for languages with small quantities of data and potentially large vocabularies. Hence our goal is not full automation, but rather to make a practical and effective workflow that integrates machine learning technologies.

AB - Machine learning has revolutionized speech technologies for major world languages, but these technologies have generally not been available for the roughly 4,000 languages with populations of fewer than 10,000 speakers. This paper describes the development of ELPIS, a pipeline which language documentation workers with minimal computational experience can use to build their own speech recognition models, resulting in models being built for 16 languages from the Asia-Pacific region. ELPIS puts machine learning speech technologies within reach of people working with languages with scarce data, in a scalable way. This is impactful since it enables language communities to cross the digital divide, and speeds up language documentation. Complete automation of the process is not feasible for languages with small quantities of data and potentially large vocabularies. Hence our goal is not full automation, but rather to make a practical and effective workflow that integrates machine learning technologies.

KW - automatic speech recognition

KW - endangered languages

UR - https://hdl.handle.net/1959.7/uws:56189

U2 - 10.21437/SLTU.2018-43

DO - 10.21437/SLTU.2018-43

M3 - Conference Paper

SP - 205

EP - 209

BT - Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India

PB - International Speech Communication Association

T2 - International Workshop on Spoken Language Technologies for Under-Resourced Languages

Y2 - 1 January 2018

ER -

Foley B, Arnold J, Coto-Solano R, Durantin G, Ellison TM, van Esch D et al. Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS). In Proceedings of the 6th International Workshop on Spoken Language Technologies for Under-Resourced Languages, 29-31 August 2018, Gurugram, India. International Speech Communication Association. 2018. p. 205-209 doi: 10.21437/SLTU.2018-43

Building speech recognition systems for language documentation : the CoEDL Endangered Language Pipeline and Inference System (ELPIS)

Abstract

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this