24th MEB Conference - Budapest, 24.04.2026

Ensuring the authenticity of student identity in online education remains a topic of research to date. With the rapid development of Internet-based distance learning system, the demand for learner identity verification methods is increasing day by day. This study explores how voice-based age and gender recognition technologies can be applied to improve the authority of online learning and examination platforms. We use deep speaker embedding technology to extract and analyze speaker identity attributes using advanced speaker verification models such as x-vector, ECAPA-TDNN and ResNet. We employ deep speaker embedding techniques to extract and analyse speaker identity attributes using state-of-the-art speaker verification models (e.g., x-vector, ECAPA-TDNN and ResNet). Using the collected speaker dataset, we evaluate the effectiveness of these models for age and gender classification, demonstrating their potential to reduce the risk of impersonation and improve the security of exam proctoring. Our findings highlight that integrating automatic speech recognition can enhance identity verification in digital educational environments while maintaining the student learning experience. This study contributes to improving biometric security in distance learning by evaluating the feasibility of voice-based authentication in distance learning.