Implementations disclose a mesh network of nearby mobile devices as a combined speaker system for audio. A method includes receiving a prompt to join a speaker mesh network as a master computing device of the speaker mesh network, the speaker mesh network comprising one or more other computing devices detected within a geographic vicinity of a computing device, generating, by the computing device acting as the master computing device, playback synchronization instructions for the one or more other computing devices, the playback synchronization instructions to cause playback of a media item to be synchronized at a playback timestamp relative to a clock time, and transmitting the generated playback synchronization instructions to the one or more other computing devices.