CPII’s Spoken Language Research on Dysarthric Speech Reconstruction awarded Championship in the SciTech Challenge 2021!
CPII’s team in spoken language research participated in the SciTech Challenge 2021 and won the championship with the entry entitled “Dysarthric Speech Reconstruction”. The goal of our work is to leverage AI-enabled speech technologies to help people with difficulty in pronouncing words due to a neuromotor disorder — a condition known as dysarthria. Dysarthria has a number of causes, including stroke, Parkinsons, brain injuries, etc. The individual needs to devote much effort to speaking and yet dysarthric speech often has reduced intelligibility for listeners. In order to overcome reduced communicative ability due to the speech impairment, our team proposed a state-of-the-art neural approach to automatically reconstruct dysarthric speech to become normal speech. This work is not only novel from the technological perspective, but it can potentially offer great enhancements towards social inclusion for dysarthric individuals. The project aligns well with the commitment of our centre to advance AI-enabled perceptual and interactive technologies for societal benefits.