Implementasi Speech Recognition Berbasis Android Dalam Optimalisasi Komunikasi Bagi Penyandang Tunarungu

Abstract

This study aims to implement Android-based Speech Recognition or Speech to Text to make it easier to communicate with deaf people, so that everyone who wants to communicate does not need to understand certain sign languages when interacting with the deaf person. The research method in developing the system uses the SDLC (System Development Life Cycle) method or often referred to as the waterfall approach. The method used to identify voice using the Vector Quantization method, which is a method for conducting learning in a supervised competitive layer, with the Vector Quantization method it can be concluded that it has a better sound accuracy and clarity value with a performance result of 93%. from the results of a survey conducted from 45 people with hearing impairment, the application that was built played a very good role in terms of the continuity of the interaction process that was carried out very smoothly and easily understood by the deaf. Keywords— Speech recognition, Speech to Text, Vector Quantization