The IPOACUSIA project aims to study and develop methods and technologies for the dissemination of Italian sign language interpreting and remote video interpreting services as well as the use of any other technology aimed at breaking down barriers to communication. The first objective is to develop a prototype network (cloud) application on which to carry out experimentation with appropriate text-to-speech and speech-to-text paradigms designed to improve communication between able-bodied people and people with hearing loss. A second goal is to develop a prototype cloud application capable of handling LIS-to-text services. One of the critical issues being addressed stems from the need to take advantage of significant computing resources for real-time management of machine learning models. In this area, the best high-performance computing architectures present at CRS4 will be tested as well as state-of-the-art options present on the cloud services market (e.g., Amazon, Google, IBM, Microsoft) will be evaluated. The operational objective of the project aims at the timely management of the various stages of the operational pipeline leading from the acquisition of the sequence of video images, to their image processing, to their simplification into a computational model useful for the automatic understanding of the gesture itself in a format suitable to be interpreted by appropriate artificial neural networks in text format. It is planned to update services developed on the basis of user feedback (LivingLab). A focus group will be created to support the development team for this purpose.
