A Comparative Assessment Of The AI-powered C-E Speech-to-Speech Translation Systems

This study makes a comparative assessment of the AI-powered C-E speech-tospeech translation systems through qualitative and quantitative methods,so as to probe into their strengths and weaknesses for the purpose of their future improvements.This study begins with literature review focusing on AI-powered speech-tospeech translation systems and quality criteria in interpreting.It first defines Artificial Intelligence,then elaborates speech-to-speech translation system and its supporting technologies,and then gives a brief introduction to Google Translate and Tencent's Mr.Translator.After a review of previous studies,the author chooses for this assessment “completeness,expression and grammar” as the quality criteria,which scholars pay most attention to.This study takes the keynote speech delivered by President Xi Jinping at the Opening Ceremony of the 1st China International Import Export as the source text,and employs Google Translate and Tencent's Mr.Translator to conduct a stimulated consecutive speech-to-speech translation.The author used the translation text released by the Xinhua News Agency as the reference in assessing the performance of two speech-to-speech translation systems.A comparative assessment of their outputs is conducted through quantitative and qualitative analyses based on the criterion of completeness,expression,and grammar.This study concludes that further improvements can be made in AI-powered SST systems by enlarging their corpora,enhancing their ability of text processing and logical analysis,and attaching greater importance to context and grammar.