Developing speech recognition technology for the Welsh language

Canolfan Bedwyr’s Language Technology Unit has been awarded a grant of £56,000 from the Welsh Government’s Technology & Welsh Language Fund and an investment from S4C’s digital fund to further develop speech recognition technology for the Welsh language.

The Unit’s drive to develop language recognition for the Welsh language was inspired by a comment from a blind man at an event to launch the Unit’s disability terminology dictionary some time ago.

Delyth Prys the Unit’s head explained:

“At first we were only involved in standardisation of terminology, but while launching our Guidelines for Using Disability Terminology back in 2001, a blind delegate at the event commented that there was a need for far more than standard terminology, and that what was truly needed was a computer that spoke Welsh. We then sought funding to develop text resources for speech, where the computer can vocalise Welsh text, and that was a great success. After this people asked to be able to speak Welsh to the computer, but that was far more difficult to deliver. This grant will be a great assistance in enabling us to develop a voice management system and a large spoken database to help us create a more sophisticated system.”

Elin Morris, Corporate & Commercial Director of S4C said:

“Welsh voice recognition technology has been discussed as a resource that would offer substantial benefits to us as a broadcaster in several ways over the coming years. As a result, S4C’s commercial arm was very pleased to be able to work with Bangor University and invest in this exciting plan. Technology develops at such a fast pace in broadcasting,  and it’s important that we recognise the opportunities  to ensure that Welsh does not get left behind, or get ousted as the natural language in Welsh speaking households. In this case, the technology will have the potential to open doors to the use of Welsh on a range of digital platforms, such as smart televisions, archiving digital content and smart telephone services. We are grateful to the Welsh Government for their financial support to aid the work in conjunction with Bangor University.”

The First Minister, Carwyn Jones, said,  “To ensure that the Welsh language flourishes in the 21st century we have to make sure that Welsh language technology and digital media are easily accessible. I’m very glad that we are able to support this development, which will lead to Welsh language voice commands having the capability to control a variety of electronic devices. We also warmly welcome the fact that research such as this is taking place in a university in Wales.”

This month sees 20 years of Terminology work at the University. Work began with a commission to standardise terminology for the National Curriculum in December 1993, when the Centre for the Standardisation of Welsh Terminology was formed at the University’s School of Education. In 2001, the Centre became part of the new Language Technologies Unit at the University’s Canolfan Bedwyr, and the new Unit also took over  over responsibility for Cysill, the University’s flagship Welsh spelling and grammar programme, which had been going since the early 1990s. By now, the Unit employs a team of 11,  including linguists, terminologists and software developers. The Unit is totally self-funding, and dependant on research grants, commissions and software licencing income to fund the work.

Professor John G Hughes, Vice-Chancellor of the University commented:

“News of the new grant is a great reason to celebrate more than twenty years of terminology and language technology work at Bangor University. The work has gone from strength to strength here, and the Language Technologies Unit are to be congratulated on their recent success.”

Publication date: 1 January 2014