We know that the bandwidth of g.711 and g.729 is 80 kbps and 24 kbps respectively. Remember that these two are enough, but it is better if you know more about them.
The enquest principle stipulates that the sound sampling frequency is 8000 times per second, 8 bits each time, and the bandwidth of the voice data is 64 Kbps (8000*8 ). A normal voice package is one packet every two frames at a time in 10 ms. In this way, 50 packets (1000/20) are generated every second ). The IP Address Header of each packet is 20 bytes, The UDP header is 8 bytes, And the RTP Header is 12 bytes, a total of 40 bytes, or 320 bits (40*8 ). 50 packages are 16000 bits (320*50) or 16 K. That is to say, the bandwidth requirement for all headers is 16 kbps. With the voice data, the total bandwidth is 80 kbps (64 + 16 ). What is the size of each package? 80 kbits/50 = 1600 bits = 200 byte (1600/8 ). The speech data size is 160 bytes (200-40, or 64 K/8/50 ). This is g.711.
G.729 adopts CompressionAlgorithmThe size of the Speech data is 20 bytes (obviously, the compression ratio is 8 to 1 compared to 160 bytes of g.711), and the packet header remains unchanged at 40 bytes, which is 60 bytes in total. The bandwidth requirement is 60*8*50 = 24000bps = 24 kbps. This is g.729.
Because the packet header is too big (twice) relative to the data load, it seems that the head is heavy, so g.729 usually compresses the packet header on the WAN circuit. The compressed header is 4 bytes or 2 bytes. Bandwidth requirements will be further reduced.