AI-driven gamified speech training for primary students: framework and evaluation

• A gamified speech system integrates AI and speech recognition technology • Gamification and AI significantly improve speech learning outcomes • A low-cost speech rate algorithm enables real-time feedback • Feynman technique improves learning when combined with gamified methods • Gamified training effectively improves student motivation Traditional public speaking education often suffers from limited learner engagement, delayed formative feedback, and a lack of interactive and adaptive training environments. This study proposes an AI-driven gamified speech learning framework (AI-GSLF), which combines real-time feedback technologies with motivational game design principles to address these issues. Based on this framework, a serious game— Strongest Speech Streamer —was developed using the Godot engine. The system integrates automatic speech recognition, sentiment analysis, and a novel speech rate detection algorithm to provide immediate feedback, helping learners adjust pacing, reduce anxiety, and enhance fluency during practice. A true experimental design was employed, involving 57 primary school students randomly assigned to either the experimental group using the gamified system or a control group following traditional methods over one month. Quantitative results showed that the experimental group demonstrated statistically significant improvements in motivation, confidence, speech accuracy, and delivery fluency. To our knowledge, few prior studies have integrated real-time AI feedback with systematic gamification for primary-level formal speech training. Findings support the potential of AI-GSLF as an effective, scalable approach to enhancing student performance and engagement in public speaking education.