People with hearing impairments rely on Sign Language as their primary means of communication. However, the widespread lack of knowledge of this system creates barriers that hinder their social and professional inclusion. With the aim of facilitating communication between hearing and non-hearing individuals, we developed a device capable of interpreting Argentine Sign Language (LSA) and translating it into spoken language. This project focuses on the recognition of short emergency related words and phrases. The system integrates Python and Arduino based platforms to capture data using sensors embedded in a glove. These data are processed by an artificial neural network (ANN) designed to recognize and classify the signs. Two ANN models were developed. Although both demonstrated high accuracy in sign classification, the second model was selected as it allows continuous inference by distinguishing between noise and valid movements. It was trained over 50 epochs reaching a loss value of \(10^{-3}\) , and an accuracy of 99,09% on the testing phase. The system was validated through two statistical analyses. One demonstrated real-time functionality with response times below five seconds, and the other evaluated robustness, reaching 92.65% accuracy when tested with external users.