The kind of medium owned by a user is stored in a service center system and a medium is determined by a request from the transmitting side in accordance with an inquiry to a receiver. Where a voice is used as a medium for the inquiry or the service, the sound piece data to which a descriptor is added and the sentence structure data for forming a single sentence by collecting the sound piece data and formed by a descriptor are stored. When the voice output request is produced, the sound piece data is obtained from the sentence structure data, thereby producing a voice of a desired sentence.