Máquinas de tradução aplicada à comunicação em tempo real para desenvolvimento distribuído de software

Abstract

The globalization allows companies to seek partnership for global project development in other countries. The domain of spoken English is considered one of the main reasons for the success of countries such as Ireland and India in global software development. However, A.T Kearney consultancy estimates that Brazil has only 5,4% of population fluent in English, and this is an important challenge that makes difficult the growth of geographically distributed software development in the country. In order to reduce the barriers posed by language differences, studies have been conducted to assess how machine translation can be used in the context of global software development in order to minimize or overcome such barriers. The goal of this research was in this context and aimed to study how speech translation can be used in global software development projects. In this research we first executed studies related to speech recognition and machine translation areas and identified available technologies for translation and speech recognition. After having identified these technologies, a simulation study was planned and executed in order to evaluate one of the identified technologies. The purpose of this study was to understand the feasibility of the available technology for speech recognition and translation for Brazilian Portuguese. The findings of the simulation indicate that Google technologies for speech recognition and translation for Brazilian Portuguese have adequate accuracy and could be used to create a complete machine translation in this language.