An investigation of AI-generated audio materials in the teaching of Hungarian medical language

Abstract

This research was conducted with the aim of providing a concise overview for instructors of Hungarian as a foreign language regarding AI-based tools—some of which are freely accessible—that can efficiently support and facilitate language teachers’ work through the creation of audio materials. We primarily examine the currently available options from the perspective of medical language instruction; however, the findings are also relevant for professionals teaching Hungarian as a foreign language at any level and in any context. The study investigated how AI performs in the following areas in the context of Hungarian medical terminology: audio materials created by instructors from specialized texts, as well as chatbots capable of autonomously generating audio content or even exchanging voice messages, optimized specifically for medical contexts and autonomous learning. Based on the results, voice generation—including samples created through voice cloning—remains an area in need of development. Applications are currently less capable of producing natural speech intonation. Overall, when examining segmental and suprasegmental features, ElevenLabs stood out among similar applications. Chatbots intended for autonomous language learning can also only be used cautiously at present; there are still shortcomings in pronunciation, while the vocabulary tends to be patient-centered and in a lay register.