Prerequisites for Advancing Automatic Speech Recognition in Breton

Proceedings of the Fifteenth Language Resources and Evaluation Conference (LREC 2026)

Abstract

We report on the extensive preliminary work of a collaborative science project aimed at developing Automatic Speech Recognition (ASR) for a minoritized European language: Breton. Hoping to help similar initiatives for other languages and communities, we present the methodology we developed for this specific ecosystem, with an estimate of the material and immaterial resources we used. Our approach is grounded in the needs and resources of the community formed by the end-users of digital development. Our multidisciplinary scientific collaboration involves linguists and speakers embedded in the academic and linguistic community, and computer scientists.