A Robin Hood approach to forced alignment: English-trained algorithms and their use on Australian languages

Sarah Babinski, Rikker Dockum, J. Hunter Craft, Anelisa Fergus, Dolly Goldenberg, Claire Bowern

Abstract


Forced alignment automatically aligns audio recordings of spoken language with transcripts at the segment level, greatly reducing the time required to prepare data for phonetic analysis. However, existing algorithms are mostly trained on a few well-documented languages. We test the performance of three algorithms against manually aligned data. For at least some tasks, unsupervised alignment (either based on English or trained from a small corpus) is sufficiently reliable for it to be used on legacy data for low-resource languages. Descriptive phonetic work on vowel inventories and prosody can be accurately captured by automatic alignment with minimal training data. Consonants provided significantly more challenges for forced alignment.

Keywords


Australian languages; phonetics; forced alignment; language documentation; Yidiny

Full Text:

PDF


DOI: http://dx.doi.org/10.3765/plsa.v4i1.4468

Copyright (c) 2019 Sarah Babinski, Rikker Dockum, J. Hunter Craft, Anelisa Fergus, Dolly Goldenberg, Claire Bowern

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Donate to the Open-Access Fund of the LSA

Linguistic Society of America


Advancing the Scientific Study of Language

ISSN (online): 2473-8689

This publication is made available for free to readers and with no charge to authors thanks in part to your continuing LSA membership and your donations to the open access fund.

The LSA is pleased to announce that Steve Anderson, former LSA President, has offered to match all donations made to the Society’s open access fund, up to $10,000, between now and the end of 2019. more ...