CUHK Research: Changing the world

Hope for local dysarthric patients Although databases for dysarthric patients are plentiful in the English language, no publicly available equivalent previously existed for Cantonese. The professors, together with a few other colleagues in CUHK, remedied this by carefully curating a selection of stimuli, the length of which range from single words to paragraphs and even full conversations in Cantonese. This allows them to gather data on the enunciation of certain words and vowels by dysarthric patients, as well as the rhythm of sentences they utter. As of 2021, some 27 dysarthric Cantonese speakers have participated in the project, generating up to 34 hours of data - a significant amount when compared to existing databases for other languages. They have also been able to adjust their system so that it automatically translates dysarthric speech into normal speech, a predictive process on which Professor Meng has brought to bear her work in AI. Together, the two Professor Meng is also Director of CUHK’s Centre for Perceptual and Interactive Intelligence, which researches speech and language intelligence and visual intelligence for the benefit of the sectors of healthcare, urban services and reindustrialisation. Neurological diseases can make communication by speech a struggle for many. When the muscles needed for speech become weak or hard to control, those affected begin to slur their words in a disorder called dysarthria. Treatment is difficult because, outside of Hong Kong, there is no body of relevant research in the Cantonese language to draw upon. Professor Helen Meng Mei-ling and Professor Patrick Chun Man Wong of CUHK have been working together for more than a decade on speech and language processing. While Professor Meng deals with the digital and AI-related aspects of their projects, Professor Wong brings neuroscience and linguistics know-how in a cross-disciplinary approach to the challenge. Their collaborative efforts have resulted in a variety of projects that cater to what they describe as “development across the lifespan.” The professors began by developing artificial intelligence models to recognise symptoms of dysarthria from speech samples, but have since broadened the scope of their research, hoping to distinguish signs of cognitive decline before patients begin exhibiting symptoms. 16