The present invention relates to method and apparatus for providing an arbitrary sound chosen by a called subscriber instead of ringback tone to a calling subscriber. In this present invention, if a call is received, a terminating exchanger requests a trunk connection to a sound db server based on the first information on whether or not to replace a ringback tone and the second information on a route to the sound db server that are received from a home location register (HLR), and provides a called subscriber identification for the sound server. Then, the sound db server searches its db for a sound specified by the called, and provides the found sound for the caller instead of a conventional ringback tone via the terminating exchanger connected through a trunk. Through this sequential procedure of network elements, a caller can hear a sound specified by a called instead of a dry ringback tone.