A system for composing multimedia signals from various source signals, such as still-frame video signals, full motion video signals and audio signals includes a plurality of composing units, each of which is capable of composing a multimedia signal from selected source signals. One of such composing units is assigned for each multimedia signal to be composed. Each composing unit accepts parameters specifying various properties for each element of the multimedia signal, such as size, position and precedence for each video element and gain for each audio element. The composing units can include decompressors for decompressing source signals furnished in compressed form and a compressor to compress the multimedia signal to be transmitted to a user. The source signals can be received from a packet network and the resulting multimedia signals transmitted to the respective users via the same packet network.