http://www.microvoip.com/technology/thh24.htm
The delay is the time [5], in general, the end-to-end delay consists of the following four parts: the end-to-end delay is composed of the following four parts:
(1) The transmission delay: the distance from one end to the other end through the network, by the signal through the speed and propagation distance from the signal
(2) Transfer delay: refers to the time of all network devices on the network path
(3) Package Transformation Delay: Refers to the time when the encoder is digitally transformed
(4) Jitter bust-delay: refers to the time of the receiving end to maintain one or more received data packets, is used to overcome the change of the data packet arrival time, that is, overcoming the delay generated by jitter. The jitter buffer typically includes two stages, the first stage is uncompressed, and if the voice data is compressed, it must be decompressed here; the second stage is transmitted to the TDM stream.
The latency can cause a blank of the speech session process, bringing voice deformation and interruption of the session, that is, the increase in delays has reduced the R value. The length of the delay starts to be perceived between 100 to 200 ms, making the session unnatural. The recommended delay upper limit is 150ms. If the delay reaches 200ms, there is a serious session interrupt.
Jitter is also called a delay, refers to the time difference between all sent packets arrive during a VoIP call. When a packet is sent, the sender adds a timestamp on the RTP packet; when the other end is received, the receiving end also adds another timestamp; calculate the two time stamps to get the path of this packet. time. There is a jitter if there is a different path time in a call. In the video application, the jitter represents the image flash, and in the telephone call, its performance is similar to the effect of the packet loss, and some words are unclear or errors. The size of the jitter depends on the degree of difference between the path time of the packet, the greater the degree of difference, the greater the jitter. If the jitter value is greater than 50 ms, the average subjective value MOS value is reduced, and it is considered to be extremely poor voice quality. By increasing the number of jitter buffers, the impact of jitter can be effectively reduced, but the jitter buffer of one millisecond is increased, the corresponding network delay is added.
