🧠 AI Ed Wiki

📄 Research Article

A Semi-Automated System for Generating Dialogue-Based TTS Lessons Using Large Language Models: An Exploratory Study of Educational Potential

◐ medium · 2026-07-15llm, generative-ai, personalized-learning, scaffolding, active-learning, pedagogical-llm-training

Gendo Kumoi, Fumie Watanabe, Tota Suko, Takashi Ishida, et al. (2026) - arXiv preprint (IEEE). arXiv preprint.

Key Findings

  • Proposes a human-in-the-loop LLM+TTS pipeline that generates Expert-Novice dialogue lessons, augmenting rather than replacing educators - see Generative AI.
  • Three-stage workflow (LLM slide/narration generation -> educator review -> automated audiovisual integration) parallels Scaffolding design.
  • Builds on cognitive apprenticeship theory to structure dialogue narration, supporting Active Learning.
  • A quasi-experiment with 245 first-year high school students suggests educational potential of dialogue-based lessons.
  • Relates to Personalized Learning and prior AI Generated Slides Student Perception work on AI-generated instructional material.
  • Informs Pedagogical LLM Training by showing how LLMs can be steered toward pedagogically sound lesson generation.

    • Connected Concepts

  • Generative AI
  • Scaffolding
  • Active Learning
  • Personalized Learning
  • Pedagogical LLM Training

    • Connected Articles

  • AI Generated Slides Student Perception

    • Citation

    Kumoi, G., Watanabe, F., Suko, T., Ishida, T., et al. (2026). A Semi-Automated System for Generating Dialogue-Based TTS Lessons Using Large Language Models: An Exploratory Study of Educational Potential. arXiv:2607.12235.