Strategies for Enhancing English Listening Skills in Low-Resource Contexts Using Generative AI
Huaqiao Zhou · 2025
Computational advancements in Generative Artificial Intelligence (AI) have revolutionized applications in education, particularly in addressing challenges faced in lowresource contexts. This study focuses on leveraging cutting-edge AI technologies, including text-to-speech synthesis, conversational AI, and AI-driven performance analytics, to enhance English listening skills. The proposed framework integrates advanced models-such as Transformers, diffusion models, and WaveNet-to generate high-quality and personalized audio content. Optimization techniques, including model quantization, ensure scalability and efficiency on lowpower devices, making the solution suitable for resourceconstrained environments. Experimental validation was conducted in rural schools with limited technological infrastructure. The system achieved exceptional performance, with audio quality metrics scoring a mean opinion score (MOS) of $4.5 / 5$ and a significant signal-to-noise ratio (SNR) improvement of $15 \%$. System responsiveness, with an average latency of less than 200 ms, supported real-time interactions. Additionally, user studies with 50 participants demonstrated an $18 \%$ improvement in standardized listening test scores after two weeks of system use, emphasizing the framework’s educational effectiveness. By addressing the technical and pedagogical challenges of low-resource settings, this study demonstrates the potential of Generative AI to democratize access to personalized and adaptive language learning tools. The findings establish a scalable foundation for integrating AIdriven technologies into education, paving the way for inclusive and impactful learning solutions.