Disclosed are a music playing method and device. A user device is used to control another user device, to realize a music on hold (playing music in a call-hold duration) function, which reduces interaction with a SIP server, thereby reducing load of the SIP server. Softswitch (i.e., switching SDP) is established between a second user device in need of the music on hold function and a media resource server by a first user device, to realize a P2P media link between the second user device and the media resource server, which saves broadband resources of the SIP server. Also, in a situation where the media resource server needs to be deployed in another network segment (i.e., the media resource server and the second user device are in two network segments), the P2P media link can avoid configuring an access permission to a router, thus being convenient for network environment deployment.