Research Seminar on Speech Technologies for Under-Resourced Languages

17 February 2026

On 17 February 2026, Lo-Rig researcher Michael Bayona gave a talk titled “Speech Technologies for Under-Resourced Languages: From Development to Application” as part of the Trinity College Dublin Linguistics Research Seminar series.

The talk provided an overview of how speech technologies such as automatic speech recognition and forced alignment can be developed and adapted in under-resourced language settings. Drawing on previous work with Philippine languages, it discussed how speech and language data can support downstream applications such as linguistic analysis, closed captioning, reading tutors, and educational technologies.

The talk also considered methodological issues that remain important for under-resourced language technology development, including limited data availability, speaker variability, pronunciation dictionary development, model adaptation, and the use of newer self-supervised learning approaches.

This addressed questions closely connected to the project's work on AI-assisted workflows for language documentation, particularly the use of speech processing tools to support transcription, analysis, and language technology development.

More information about the seminar is available on the Trinity College Dublin website:

Speech Technologies for Under-Resourced Languages: From Development to Application