[asterisk-users] G.729 codec between avaya and asterisk

[Anselm Martin Hoffmeister]

Am Dienstag, den 23.10.2007, 22:21 -0700 schrieb satish patel:
> there is no special requiremnt to use g.729 but day to day my sip
> client incressing thats why some time i got breaking voice or voice
> quality not much better i think in LAN there is lots of brodcat on
> lan 

If your LAN is congested and a lot of single packet delay happens, you
should improve the LAN. You cannot run a LAN at 99% saturation with
VoIP, it will just not work, with packet drop rates and delays making
phone calls more of a earth-to-moon radio experience ("Houston *crackle*
*crackle* have *crackle* problem").

If _all_ that traffic is VoIP, G729 might help a bit, but I would not
expect it to get around all your bandwidth problems. Try to improve the
network first.

One interesting aspect of g729 might be that your sip client phones that
live behind a DSL line might profit from the smaller bandwidth
requirement on their side.

> if i purches g.729 transcoder license for asterisk to convert g.729 to
> g.711 then  it will work or not

I _think_ it will work (btw this is, as of some website I found, the
"main revenue stream" of Digium, so they will be interested in having it
working). Others with real-world experience could tell you.

> but why i need codec on trunk ????

Codec stands for coding-decoding (or something similar). If you imagine
the "original signal" as voice and sound, meaning variations in air
pressure around the membrane of the telephone handpiece microphone, then
every digital representation is a kind of "coding". This even refers to
8-bit-wave, which is the most obvious way of encoding: It merely writes
down the voltage level at the microphone input in the range -128 to +127
(IIRC, correct me if I am wrong). Accordingly, 16-bit wave has the
higher precision of -32768 to +32767.

G711 is - again, if I remember correctly - an adaptation of these bytes
to a logarithmic scales, bearing in mind the idea that small changes in
the higher ranges are treated differently from small changes in the
near-0-region. Something like the fiction bytestream value 0 1 2 3
representing the scale 0 4 6 7 of microphone values, instead of linear
data. Please research this yourself if you are interested in details.

G711 is the standard (and usually, the only available) codec for
ISDN/T1/E1... Europeans and US Americans established two different kinds
of G711 (µ-law and a-law) which seem to be functionally similar.

BR
Anselm