A process in which a synthesizer that is being required to play more notes than it has polyphony switches off some currently sounding voices (typically those that have been sounding the longest or are at the lowest amplitude) in order to assign them to play new voices. See dynamic voice allocation.