A voice service system which includes a database storing information and is connected with a plurality of terminals by a communication network, comprising a call connection unit for transmitting-receiving a signal, on a voice service to/from each of the terminals, a voice processing unit for interpreting a voice instruction received from the terminal and converting an internally-created voice service document into a voice message, a recognition management unit for extracting a grammar, which is expected to be spoken by a user, from the database, a document management unit for creating the voice service document including the extracted grammar as a voice anchor, and a control unit for controlling information related to the voice anchor matching with the voice instruction to be provided.