Provided are a method and a device for processing voice information, a storage medium and an electronic apparatus. The method comprises: searching whether there is voice information matching first question sentence voice information inputted by a user (S202); in a case that a searching result is no, performing semantic analysis on the first question sentence voice information, and generating second question sentence voice information according to the first question sentence voice information and a semantic analysis result of the first question sentence voice information (S204); searching the second question sentence voice information, and determining, according to a searching result, voice information of a rhetorical question sentence corresponding to the second question sentence voice information (S206); and determining, according to an answer of the user to the rhetorical question sentence, question sentence voice information to be fed back to the user (S208). The present invention solves the problem in the related art of limitation in recovering omission-type semantic loss on the basis of a semantic understanding result of a knowledge graph, thereby improving the user experience.