<![CDATA[Artificial intelligence (AI)-assisted speaking tools have increasingly been integrated into English for Academic Purposes (EAP) instruction to support oral communication. However, limited research has examined how AI-mediated speaking practice influences oral fluency while simultaneously addressing speaking anxiety and learner autonomy among Khmer EFL university students. This study investigates the effects of AI-mediated speaking practice on the oral fluency, speaking anxiety, and learner autonomy of ten Khmer EFL university students. Using a qualitative case study with embedded quantitative fluency analysis, participants completed nine AI-assisted speaking sessions over a three-week intervention using the EAP Talk platform. Oral fluency was assessed through Praat acoustic analysis focusing on speaking ratio, repair phenomena, and pause placement, while reflective questionnaires and semi-structured interviews explored learners' perceptions of anxiety and autonomy. The findings indicate that participants demonstrated longer speaking duration and fewer repair phenomena following the intervention, while pause placement shifted toward more meaningful syntactic boundaries. Questionnaire and interview data also revealed reduced speaking anxiety and improved learner autonomy during AI-mediated practice. These findings suggest that AI-assisted speaking practice can provide supportive opportunities for developing oral fluency while fostering greater confidence and self-directed learning in EAP contexts. Nevertheless, the findings should be interpreted cautiously because of the small sample size and the absence of a comparison group. Future studies should employ larger samples, longer intervention periods, and experimental designs to provide stronger evidence regarding the effectiveness of AI-assisted speaking practice.]]>
