Automatic Speech Recognition for Supporting Endangered Language Documentation
Date
2021-11
Contributor
Advisor
Department
Instructor
Depositor
Speaker
Researcher
Consultant
Interviewer
Narrator
Transcriber
Annotator
Journal Title
Journal ISSN
Volume Title
Publisher
University of Hawaii Press
Volume
15
Number/Issue
Starting Page
491
Ending Page
513
Alternative Title
Abstract
Generating accurate word-level transcripts of recorded speech for language documentation is difficult and time-consuming, even for skilled speakers of the target language. Automatic speech recognition (ASR) has the potential to streamline transcription efforts for endangered language documentation, but the practical utility of ASR for this purpose has not been fully explored. In this paper, we present results of a study in which both linguists and community members, with varying levels of language proficiency, transcribe audio recordings of an endangered language under timed conditions with and without the assistance of ASR. We find that both time-to-transcribe and transcription error rates are significantly reduced when correcting ASR for language learners of all levels. Despite these improvements, most community members in our study express a preference for unassisted transcription, highlighting the need for developers to directly engage with stakeholders when designing and deploying technologies for supporting language documentation.
Description
Keywords
Automatic speech recognition, Endangered languages, Transcription
Citation
Prud'hommeaux, Emily, Robbie Jimerson, Richard Hatcher, Karin Michelson. 2021. Automatic Speech Recognition for Supporting Endangered Language Documentation. Language Documentation & Conservation 15: 491-513.
Extent
23
Format
Geographic Location
Time Period
Related To
Related To (URI)
Table of Contents
Rights
Creative Commons Attribution-NonCommercial 4.0 International
Attribution-NonCommercial 3.0 United States
Attribution-NonCommercial 3.0 United States
Rights Holder
Local Contexts
Email libraryada-l@lists.hawaii.edu if you need this content in ADA-compliant format.