Semiautomatic Speech Alignment for Under-Resourced Languages

Juho Leinonen, Niko Partanen, Sami Virpioja, Mikko Kurimo

Research output: Chapter in Book/Report/Conference proceedingConference article in proceedingsScientificpeer-review

41 Downloads (Pure)

Abstract

Cross-language forced alignment is a solution for linguists who create speech corpora for very low-resource languages. However, cross-language is an additional challenge making a complex task, forced alignment, even more difficult. We study how linguists can impart domain expertise to the tasks to increase the performance of automatic forced aligners while keeping the time effort still lower than with manual forced alignment. First, we show that speech recognizers have a clear bias in starting the word later than a human annotator, which results in micro-pauses in the results that do not exist in manual alignments, and study which is the best way to automatically remove these silences. Second, we ask the linguists to simplify the task by splitting long interview audios into shorter lengths by providing some manually aligned segments and evaluating the results of this process. We also study how correlated source language performance is to target language performance, since often it is an easier task to find a better source model than to adapt to the target language.
Original languageEnglish
Title of host publicationProceedings of the Workshop on Resources and Technologies for Indigenous, Endangered and Lesser-resourced Languages in Eurasia (EURALI)
PublisherEuropean Language Resources Association (ELRA)
Pages17-21
Number of pages5
ISBN (Print)978-2-493814-07-4
Publication statusPublished - 2022
MoE publication typeA4 Conference publication
EventWorkshop on Resources and Technologies for Indigenous, Endangered and Lesser-resourced Languages in Eurasia - Marseille, France
Duration: 20 Jun 202220 Jun 2022

Workshop

WorkshopWorkshop on Resources and Technologies for Indigenous, Endangered and Lesser-resourced Languages in Eurasia
Abbreviated titleEURALI
Country/TerritoryFrance
CityMarseille
Period20/06/202220/06/2022

Fingerprint

Dive into the research topics of 'Semiautomatic Speech Alignment for Under-Resourced Languages'. Together they form a unique fingerprint.

Cite this