A SIP enabled softswitch typically uses a separate media server for music on hold (MoH). Different extensions or SIP addresses are be assigned for different MoH sources on the media server.
Take the example where two parties (A and B) are talking on SIP phones. When party A presses the HOLD button, a SIP re-INVITE message is sent to the softswitch indicating that the call should be placed on hold (e.g., party B should hear music). The softswitch then sends a re-INVITE message to party B which indicates it should establish a new call with the MoH source. So, SIP is used to set up the call so the audio stream for the MoH can be played.
This was first published in September 2005