Methods and systems for synchronizing audio layers is disclosed. In one aspect, a method of synchronizing audio layers to form an audio composition includes receiving, from a first computing device, at an audio synchronization system, a first audio layer having a first length. The audio synchronization system identifies a portion of a second audio layer with which the received first audio layer is to be synchronized. The audio synchronization system adjusts the first length to match a second length of the identified portion of the second audio layer by altering the tempo of the first audio layer. The audio synchronization system trims a recording signal delay from a beginning portion of the first audio layer and combines the first audio layer and the second audio layer to form an audio composition. In some implementations, the second audio layer includes a reference audio layer.