Sistema traductor a lenguaje de Señas para reducir la dificultad de comunicacion en personas con discapacidad auditiva en Ecuador

Abstract

According to the National Council for Disability Equality (CONADIS) in Ecuador, it is estimated that approximately 66,538 Ecuadorians suffer from some hearing disability. In response to this problem, a notable increase in Sign Language translation tools has been observed. This scientific article presents a web application for text translation into the Sign Language of Ecuador (LSEC). The logic of the web application, developed in Python with FastAPI, and the graphical user interface, implemented in Angular, follow the Model View Controller (MVC) architecture. The controller employs non-deterministic finite automata (AFND) and regular expressions for translation. Users can enter text using a standard keyboard and receive visual translations in the form of animations representing LSEC signs. The performance evaluation showed an average response of 0.28 seconds per sentence translation request during high-demand hours and 6% failures per request per second. The usability and accessibility of the tool were evaluated in collaboration with the National Federation of Ecuadorians with Physical Disabilities (FENEDIF), which allowed improvement and correction of errors in the visual representation of LSEC signs and the adoption of a more appropriate color palette. In the evaluation of the web application for text translation to LSEC, 75% was obtained in the categories of “very easy” or “moderately easy” for usability and 77.77% in the categories of “good” and “excellent” for accessibility.