NWU speech research group joins engineering faculty - 31 January 2018

The Multilingual Speech Technology research group (MuST) within the North-West University is joining forces with the NWU Faculty of Engineering in 2018. This research group focuses its efforts on the creation and use of speech technologies and the study of the underlying machine learning principles. The group has a specific interest in deep learning techniques and their application in various domains, including speech and language processing.

The group has a long history of developing speech technologies for the lesser-resourced languages of the world. Although the applications it develops focus mostly on South African languages, the group has worked on 26 different lesser-resourced languages as part of the US IARPA Babel project, including diverse languages such as Amharic, Dholuo, Igbo, Javanese, Kurdish and Mongolian.

In 2016, MuST collaborated with Google to create new South African voices speaking four South African languages: Sesotho, isiXhosa, English and Afrikaans. It also recently completed the development work on a speech transcription platform for South African languages. The project was executed in close consultation with the reporting unit of National Parliament, and includes workflow management and various text and speech processing features, including automatic speech recognition.

When considering the current engineering research projects, MuST was a natural fit with the faculty of engineering. The senior research staff in MuST all have an engineering background, with formal PhD qualifications in electronic and computer engineering. As MuST delivers postgraduate rather than undergraduate students, the partnership with the faculty of engineering provides them with access to students with exposure to the appropriate undergraduate syllabus. In turn, the faculty gets to collaborate on high-profile research projects, new sources of research topics and funding, and a unique network of collaborators.

MuST is the NWU node of the Centre for Artificial Research (CAIR), a national research network hosted by the CSIR and mostly funded by the Department of Science and Technology (DST). As such MuST has access to attractive bursaries for post-graduate studies in the field, and an impressive network of scientists working on various aspects of AI. According to Prof Marelie Davel, director of MuST, the plan for 2018 is to intensify the machine learning aspect of their research focus.

For more information contact Professor Marelie Davel, MuST, +27 (0)28 312 1907, marelie.davel@nwu.ac.za