This mixed‑methods study investigates the efficacy of an emotionally intelligent (EI) AI platform, Amazon Alexa, in enhancing English speaking skills among EFL learners. Forty Iranian high‑school students (aged 16–18) were randomly assigned to an experimental group (n = 20) that received EI‑driven, real‑time adaptive feedback via Alexa, or a control group (n = 20) that used a standard ChatGPT‑3.5 interface. Following an eight‑session intervention, quantitative data from a researcher‑designed Speaking Assessment System and a Perception Questionnaire were analyzed alongside qualitative insights from classroom observations and semi‑structured interviews (n = 15). A one‑way between‑subjects ANCOVA, controlling for pre‑test scores, revealed a statistically significant main effect of the EI‑enhanced intervention on post‑test speaking proficiency, F(1, 37) = 41.27, p < .05, η² = .197. The Alexa system demonstrated 94% accuracy in real‑time emotion detection, enabling personalized feedback that directly addressed learners’ affective states. Results indicate that the integration of emotionally intelligent AI not only improves speaking performance but also substantially reduces speaking anxiety by simultaneously targeting cognitive and affective dimensions of language acquisition. The findings underscore the scalability and cross‑cultural applicability of EI‑aware platforms, positioning them as transformative tools for advancing global language‑education practices.