Abstract Public speaking is a vital skill for academic and professional success, yet numerous learners struggle with glossophobic bia, limited vocabulary, and weak delivery. Being training approaches, similar to practices or peer evaluation, give limited feedback and aren't scalable. This paper introduces \textit{Lingo-Test},an AI-powered platform designed to enhance spoken English through real-time, multimodal feedback. The system integrates Automatic Speech Recognition(ASR), Natural Language Processing(NLP), aspect discovery, and facial expression analysis to estimate both verbal and non-verbal performance. Unlike conventional tools that concentrate only on pronunciation or alphabet, \textit{Lingo-Test} generates a compound confidence score and individualized recommendations covering ignorance, tone variation, vocabulary precarious, and eye contact. The architecture includes modular factors for speech processing, sentiment analysis, and feedback visualization, making it scalable for academic and professional operations. An pilot study with learners indicated measurable advancements in ignorance, confidence, and non-verbal delivery. These results punctuate the eventuality of AI- driven multimodal feedback systems to reduce anxiety, strengthen tone mindfulness, and ameliorate communication chops, situating\textit{Lingo-Test} as a practical result for education and training