Abstract: A virtual scene interactive language learning system and method adopts a server/user end architecture to simulate a real language environment to enable users to learn language in selected scenes. Current scene conditions are confirmed by data exchange between the user end and the server end. A cycle editing is performed for scene elements according to the scene conditions. The edited interfaces are presented to the user end. Voice data input at the user end are received to make corresponding responses at the user end so that users can learn the language deftly.