Towards Computer Aided Transformation of Spoken Text to Cued Speech for Speech Therapy

The long-term goal of our interdisciplinary research is to prepare software for AR glasses to automatically convert spoken text into cued speech particles, and then, to visualize the sequence of hand positions and locations near the face according to the Cued Speech Method principles, which will accompany natural mouth movements during speaking. In the paper, we consider a problem of computer aided transformation of spoken text to cued speech. We describe formally a procedure that has to be made for transformation, and next, we show results of tests of several approaches (using large language models and a direct implementation of rules) to automatic transformations as well as expert transformations as reference.