Quality evaluation of iFLYTEK's interpretation output: a case study of Jack Ma and Elon Musk's debate at WAIC 2019

Abstract

Interpreting software technology has grown in popularity over the last decade, particularly at formal conferences and meetings. Machine interpretation, which features AI technology and the neural network, has sparked a heated controversy, and public perceptions of the machine-generated interpretation quality vary. Research on in-depth analysis and evaluation of machine interpretation, on the other hand, is far from adequate. With the aim of acquiring objective knowledge of interpreting software, this dissertation studies the application of iFLYTEK simultaneous interpretation in a conversation between Jack Ma and Elon Musk at the 2019 World Artificial Intelligence Conference. In the case study, the author conducted a thorough analysis based on established qualitative and quantitative evaluation standards and classified typical mistakes in the interpretation output into two categories: speech recognition mistakes and machine translation mistakes. Suggestions for future enhancements to interpreting software are proposed in light of the interpretation errors and overall quality evaluation.