![]() |
![]() |
Each IP frame carrying voice data has the following overhead while on an ethernet LAN link:
Ethernet Header |
14 bytes |
IP Header |
20 bytes |
UDP Header |
8 bytes |
RTP Header |
12 bytes |
Ethernet Trailer |
4 bytes |
Total |
58 bytes |
If layer 2 (the data link layer) is to be included in the header calculation, increase this result by 8 bytes. If 802.1 p/Q fields are included, increase by another 4 bytes. The one-way voice data rate for each of the codecs is given in kilobits per second (kbps). To convert this to bytes per 30 ms, multiply by 30/8. Add the result to the header/trailer overhead (58) to get the total frame size. For example, a G.711 64 kbps voice sample is 64 x 30/8 = 240 bytes. Dividing by the compression ratios for the other codecs yields their data bytes per 30 ms frame. Adding on 58 bytes for the header/trailer yields the total 30 ms frame size.
Codec |
Compression Ratio |
Data Bytes' 30 ms Frame |
Total Frame Size (bytes) |
---|---|---|---|
G.711 |
1 |
240 |
298 |
G.729 |
8 |
30 |
88 |
G.723a |
10 |
24 |
82 |
G.723b |
12 |
20 |
78 |
To calculate the data rate of these frames, multiply by 8/30. For example, the G.711 result is
To determine the data rates for other frame sizes, convert the 30 ms data bytes per frame to the appropriate result, add on the header, and convert to kbps. For example, to calculate the G.711 rate for 10 ms frames, divide the 30 ms result (240) by 3, add 58, and multiply by 8/10. The result is 110.4 kbps. G.711 and G.729 frames are multiples of 10 ms. G.723 frames are multiples of 30 ms. For voice quality reasons, frames should not exceed 60 ms.
![]() |
![]() |