An AI-Driven Pronunciation Feedback and Exercise Generation System using Real-Time Phoneme Analysis for Language Learning

Sai Vishaal Saibaskar, S. Rubin Bose, P. Taroon, R. S. Gaayathri, Sebastiana Salerno, Anurag Hazarika, Targyn A. Nauryz, Hernández Hernández Department of Data Science, University of Wisconsin, Madison, Wisconsin, United States of America. School of Computer Science and Engineering, SRM Institute of Science and Technology, Ramapuram, Chennai, Tamil Nadu, India. Department of Science, Educandato State Maria Adelaide, Palermo, Sicilia, Italy. Department of Management, Global Interfaith University, Lewes, Delaware, United States of America. School of Digital Technologies, Narxoz University, Almaty, Kazakhstan. Department of Engineering and Technology, Autonomous University of Tlaxcala, Tlaxcala, Mexico.

Abstract: In this paper, researchers present a system for generating pronunciation feedback and exercises that aims to improve language acquisition through real-time phoneme analysis. The proposed system is built on advanced speech recognition and Generative Artificial Intelligence (GenAI) techniques to assess learners’ pronunciation accuracy and automatically generate personalized exercises for specific phonetic difficulties. The framework also features a straightforward interface that allows users to interact with the system and receive instant feedback on their speech production. The system analyses the phonemic structure of learners’ utterances, detects differences between target and actual pronunciations, and thus offers detailed insights into pronunciation error patterns. Based on these observations, specific exercises are generated on the fly to reinforce problematic phonemes and improve speech intelligibility. The use of GenAI also enables adaptive learning, which can adjust the difficulty and focus of exercises based on the learner’s progress and performance. Experimental results demonstrate that the proposed system may significantly improve pronunciation proficiency through continuous, individualized, and data-driven practice, making language learning more efficient and accessible. This paper discusses the system architecture, the implementation approach and its potential impact on modern language education and intelligent learning environments.

---

Keywords: Pronunciation Feedback; Phoneme Analysis; Personalized Exercises; Generative Artificial Intelligence (GenAI); Speech Recognition; Speech Production; Language Learning.

---

Received on: 23/08/2025Revised on: 04/11/2025Accepted on: 25/11/2025Published on: 03/03/2026

---

---