Jejueo Datasets for Machine Translation and Speech Synthesis

TL;DR

This paper introduces two new datasets for Jejueo, a critically endangered language, enabling advancements in machine translation and speech synthesis to aid its revitalization.

Contribution

The creation of the first large-scale Jejueo datasets for machine translation and speech synthesis, facilitating computational language preservation efforts.

Findings

Successful development of neural translation systems

High-quality speech synthesis results

Datasets publicly available for research

Abstract

Jejueo was classified as critically endangered by UNESCO in 2010. Although diverse efforts to revitalize it have been made, there have been few computational approaches. Motivated by this, we construct two new Jejueo datasets: Jejueo Interview Transcripts (JIT) and Jejueo Single Speaker Speech (JSS). The JIT dataset is a parallel corpus containing 170k+ Jejueo-Korean sentences, and the JSS dataset consists of 10k high-quality audio files recorded by a native Jejueo speaker and a transcript file. Subsequently, we build neural systems of machine translation and speech synthesis using them. All resources are publicly available via our GitHub repository. We hope that these datasets will attract interest of both language and machine learning communities.