Abstract: FR is a packet-orientated technology (with the packets of various lengths); it is impossible to give any warranty with regard to delay values, which are critical during an audio information transfer. In event of any node overfilling or fault, the packets containing the audio information can be redirected (in case of a node it can be done using technology of the commutable permanent virtual channel) that can lead to delay increase. Such delays can be escaped only by means of the traffic effective control and the entire network reliability increase. As the FR protocol assumes the various lengths of packets, the data packets buffering in the FR switches also can cause increase in delay values scatter (the frames with audio information have to wait until the longer frames are transferred to communication channel). It is possible to eliminate delay in the switches by means of the priorities application to the frames and the longer packets defragmentation. The both above mentioned methods require standardized technologies. One more potential delay source is a time necessary to fill the frame with the audio information. In order to achieve efficient use of the entire pass band and to escape the transmission of frames, which practically do not contain information, all the frames should contain information. Determined by duration of the frame filling process, delay is calculated as T = L/R (T is the time necessary to fill the frame by information; L is the frame length; R is the data receive rate). Delay exceeding 200 ms becomes appreciable and annoying to user. This problem can be solved by means of the frame length contraction[1].