In the Language Bank: Katri Hiovain-Asikainen

Kielipankki – The Language Bank of Finland offers a comprehensive set of resources, tools and services in a high-performance environment. Katri Hiovain-Asikainen describes her research on spoken Sámi languages and speech synthesis.

Who are you?

I am Katri Hiovain-Asikainen, working my fourth year as a speech and language technologist in the Divvun group at the Arctic University of Norway. Our group develops language and speech technology applications especially for the Sámi languages, but also for other minority languages. I am responsible for the design and implementation of speech technology projects, in which collecting different types of audio data and building speech corpora for different Sámi languages is essential.

This year, our team has published the world’s first speech synthesis for Lule Sámi, and updated North Sámi speech synthesis to modern standards. In October, we also published the world’s first South Sámi synthesis. All the software and tools that we have developed are free and easily accessible to all.

My background is in linguistics and phonetics, and I received my PhD from the University of Helsinki in autumn 2023. The topic of my dissertation was the influence of the majority languages on the spoken North Sámi language. The aim of the research was to investigate the variations of prosodic features, such as quantity and intonation, in the regional spoken varieties of North Sámi, where the contacts with the majority languages (Finnish and Norwegian) are very close and multidimensional.

What is your research topic?

Currently, I focus on the development of speech synthesis and automatic speech recognition for three Sámi languages: North, Lule and South Sámi, all of which are official languages in Norway. There is a great need for speech technology applications in the Sámi-speaking communities, as written forms of the Sámi languages are relatively new, and not all Sámi speakers have had the opportunity to learn the written language in school in the same way as the speakers of the majority languages. Speech technology enables the oral use of minority languages in new contexts: for example, as a reading assistant at school, for learning the pronunciation, as an easy-to-use tool for dyslexic or visually impaired people, and in general, even for listening to the news instead of reading. Audio books and other spoken language content are also becoming more common, allowing you to listen to books while doing something else with your hands. Today, a smart home and smart loudspeakers speak Lule Sámi in a home where the language of the family is Lule Sámi. This strengthens the role of the language and supports the revitalisation of Sámi languages at a new level.

An automatic speech recognizer, on the other hand, enables different speech interfaces, for example in the car and at home, and of course on smart devices. It will soon be possible to dictate texts in Sámi languages and, for example, to produce automatic transcriptions for old archival recordings so that researchers can make better use of them. The possibilities are endless.

The focus of my research is strongly related to speech technology, and I am currently a visiting researcher in the Phonetics and Speech Synthesis Research Group at the University of Helsinki. In collaboration with other researchers in the group, we have been working on automatic dialect recognition, where the aim is to automatically identify the speaker’s dialect based, among other things, on various prosodic features. In addition, I am interested in different methods of speech synthesis evaluation, for example, how well the speech synthesis learns to produce complex and rare prosodic features such as quantity.

How is your research related to Kielipankki?

In the Divvun group we are currently preparing various Sámi speech corpora for publication via Kielipankki. There are Sámi archive recordings in different countries, but they are relatively scattered or not necessarily processed for publication, and transcriptions are not always available. We believe that making these existing materials more accessible would help many researchers and developers of speech technologies without making new recordings.

I have also gained access to a North Sámi speech corpus (Giellagas) in Kielipankki for research purposes, and the corpus has been very useful because of its versatility, especially in the study of automatic dialect recognition. Our aim at Divvun is to make similar corpora available as soon as possible. However, in the case of indigenous and minority languages, the publication of the corpora should be treated with caution, which we respect in our work.

Recent publications

Hiovain-Asikainen, K. (2023). Prosodic change and majority language influence in spoken North Sámi varieties. Helsingin yliopisto, Humanistinen tiedekunta, Digitaalisten ihmistieteiden osasto. Helsingin yliopisto.

Kakouros, S., & Hiovain-Asikainen, K. (2023). North Sámi dialect identification with self-supervised speech models. arXiv Preprint arXiv:2305.11864. In Proceedings of the 24th INTERSPEECH Conference (pp. 5306–5310).

Pirinen, F., Moshagen, S., & Hiovain-Asikainen, K. (2023, May). GiellaLT—a stable infrastructure for Nordic minority languages and beyond. In Proceedings of the 24th Nordic Conference on Computational Linguistics (NoDaLiDa) (pp. 643-649).

Hiovain-Asikainen, K., & de la Rosa, J. (2023). Developing TTS and ASR for Lule and North Sámi languages. In Proceedings of the 2nd Annual Meeting of the Special Interest Group on Under-resourced Languages (SIGUL)

Corpora and Tools

  • Giellagas, Samples of Northern Saami
  • Borealium – tools for the small languages of the Nordic countries.

More information

The FIN-CLARIN consortium consists of a group of Finnish universities along with CSC – IT Center for Science and the Institute for the Languages of Finland (Kotus). FIN-CLARIN helps the researchers of Social Sciences and Humanities to use, refine, preserve and share their language resources. The Language Bank of Finland is the collection of services that provides the language materials and tools for the research community.