Potential of Synthetic Voice in Multilingual Educational Environments: Utilizing Generative AI in XR and Metaverse-Based International Virtual Exchange

Abstract

In this work-in-progress paper, we analyze the impact of using synthetic, natural, and text-to-speech (TTS) voices in a flipped classroom-style lecture video setting. Our study aims to give an insight on how different voice types affect students' comprehension, perception of naturalness, preferences and overall satisfaction, especially in an international and multilingual learning environment. Participants watched three versions of the same lecture, each only differing in the voice used for narration and evaluated them across various criteria. Results showed that while a natural voice remains slightly preferred, high-quality synthetic voices can perform comparability in terms of clarity and acceptability.