Testing the efficiency of voice recognition software in translation
Marko Pernarčić · 2019
Today, the translation service industry and the speech technology industry are both in significant growth, showing no indication of stopping in the near future. Speech recognition technology is slowly becoming integrated in Computer Assisted Translation (CAT) tools, which has the potential to increase productivity of the translation process. The aim of this paper is to determine whether speech recognition is a more efficient input method than typing in the translation process—assessing primarily speed and accuracy. Additionally, the goal was to establish the level of adequacy of the integration of speech recognition technology with CAT tools in its current instance. The research consisted of tests through which the respondents were tested for their typing, dictating, and translating abilities, along with speed. Two independent evaluators assessed the translations, ultimately using the average of scores and grades. We have established a sizable number of metrics and used three different nonparametric tests in order to test the hypotheses for 8 research questions. Based on the results obtained, we have provided a foundation for improving the translation process.