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

A research study proposes a semi-automated system leveraging Large Language Models (LLMs) and Text-to-Speech (TTS) technology to create dialogue-based educational lessons. This system operates with a human-in-the-loop approach, allowing educators to review LLM-generated content before automated audiovisual integration. It also introduces a novel method for generating Expert-Novice dialogue narration, aligning with cognitive apprenticeship theory, and explores its educational potential through a quasi-experiment with high school students.

Why it matters

This development represents a potential evolution in educational content creation, offering tools that could enhance learning experiences and potentially scale resource delivery. The human-in-the-loop design suggests a model for technology integration that preserves professional oversight, crucial for maintaining quality and relevance in diverse learning environments.

Key insights

• A semi-automated system has been developed for generating dialogue-based lessons using LLMs and TTS.
• The system employs a human-in-the-loop workflow, augmenting rather than replacing educators.
• The generation process involves LLM-based content generation, educator review, and automated audiovisual integration.
• A novel method for Expert-Novice dialogue narration, based on cognitive apprenticeship theory, is a key component.
• An exploratory quasi-experiment involving 245 first-year high school students is being conducted to assess educational potential.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2607.12235