Hello List:
between G723.1 and G729, what codec do you prefer?
I am just trying to understand pros and cons of both codec.
Thanks, MOhammad
-------------------------------------------------------------------- mail2web - Check your email from the web at http://mail2web.com/ .
G723: primary advantage is very low bandwidth requirements
Answering a question like that is impossible. It depends on your deployment scenario, what you are trying to achieve, the type of service you are delivering etc. The MOS value for G729 is 4.0, I believe (G711 has 4.3), with G723 well below 4.0. Thus, the perceived quality of the call will be considerable less than for 729. WAN-type deployments tend to prefer G729. g-)
Kofi Obiri-Yeboah wrote:
G723: primary advantage is very low bandwidth requirements
Hello List:
between G723.1 and G729, what codec do you prefer?
I am just trying to understand pros and cons of both codec.
Thanks, MOhammad
mail2web - Check your email from the web at http://mail2web.com/ .
Serusers mailing list serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
Serusers mailing list serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
Just to add a few more details, Greger is right to point out the quality inferiority of G.723 compared to those of G711 and G729. In fact, in most VOIP deployments, in order to quarantee interoperability, a minimum bandwidth of 128K is specified. However to reach the wider "lower bandwidth areas" most service providers are opting for G.723 which uses either 5.3 or 6.3K. At this low bandwidth transmission needs, one could literally reach "dial up modem" equipped areas. in fact most VOIP phone hardware and software are begining to specify G.723 as their default codec. Note that until the direct media connection phase of a VOIP vall setup, wide bandwidth is not required. Also note that analogue phones have a maximim bandwidth need of 3K, hence even the low quality of G.723/5.3K, compared to the average analogue phone call, is superior
the problem with BOTH g.723 and g.729 is that they are not free and thus cannot be included in free, open source sip UAs. for that reason, i don't see any reason to favor their use. the free alternatives are iLBC and speex.
-- juha
the success or failure of VOIP is dependent on inter operability. While I agree that "free" is better than "pay", unfortunately, for now, one should deploy whatever codec is more popular. remember that the sdp protocol allows for codec negotiation.
Kofi Obiri-Yeboah writes:
the success or failure of VOIP is dependent on inter operability. While I agree that "free" is better than "pay", unfortunately, for now, one should deploy whatever codec is more popular.
with that kind of philosophy, why don't you stick to windows and forget ser and this list.
-- juha
Ouch !!
Juha Heinanen wrote:
Kofi Obiri-Yeboah writes:
the success or failure of VOIP is dependent on inter operability. While I agree that "free" is better than "pay", unfortunately, for now, one should deploy whatever codec is more popular.
with that kind of philosophy, why don't you stick to windows and forget ser and this list.
-- juha
Serusers mailing list serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
.
You will be surprised to know I cut my teeth on UNIX back in the 70s. and still use UNIX (Linux debian) as my primary OS. Since VOIP is our main source of income we do what needs to be done to stay in business. check what iptel does to survive and you will once again be surprised. Open source does not imply bad business sense. Also talking about open source, how many of us really support the movement in the area where it counts - financial or otherwise. The true philisophy of open source includes financial and other means of support!
Disagree two big providers Skype, and Vonage....I'll throw in a 3rd Net2phone, also, since when did they ever interoperate with anyone, at a primary level, let alone a codec level.
VoIP will work if the VoIP providers interconnect with each other, the end points can take care of the codec negotiation, I guess it adds overhead, but the more the merrier I think, and if its "free" the better it will be, rather than the likes of cisco locking all development down to one or the other, and hiking up prices.
From a commercial point of view, all consumers want awesome quality and cheap prices, the internet has given them this notion that everything should be a $/£10 per month. If you have that, and they want the hardware subsidised, then when u have paid for codecs inserted, the business model just does not work, unless you have volumes. Vonage spends approx $110 in acquiring a customer, and it is a loss leader for them, but how many companies including those on this list have a cash corpus of $200+ million...if they do my services are for hire :-). They may not care of the cost of a codec....but I need to
If a really good codec for low bandwidth hits the market it can only work if it is free...why? simply low bandwidth is currently deployed in places where internet penertration is low...due to policy or cost , and hence adding a further additonal cost onto the hardware, will not help it spread. So keep it free, and I know some of the guys on this list....and others like it, will make it all work together :-)
Just my $0.02
Iqbal
Kofi Obiri-Yeboah wrote:
the success or failure of VOIP is dependent on inter operability. While I agree that "free" is better than "pay", unfortunately, for now, one should deploy whatever codec is more popular. remember that the sdp protocol allows for codec negotiation.
the 5.3, 6.3K are really theoretical, i dont think they include IP overheads, I used media_sessions.phtml, and looked at the actual calls per codec, and I dont think u can really get a good call without 50-70K, also most bandwidth providers (at least here in the UK) are asymmetrical, so even on a 128K, u could have problems.
Having said that I have done a nice call on xlite using ilbc on dial up.
Iqbal
Kofi Obiri-Yeboah wrote:
Just to add a few more details, Greger is right to point out the quality inferiority of G.723 compared to those of G711 and G729. In fact, in most VOIP deployments, in order to quarantee interoperability, a minimum bandwidth of 128K is specified. However to reach the wider "lower bandwidth areas" most service providers are opting for G.723 which uses either 5.3 or 6.3K. At this low bandwidth transmission needs, one could literally reach "dial up modem" equipped areas. in fact most VOIP phone hardware and software are begining to specify G.723 as their default codec. Note that until the direct media connection phase of a VOIP vall setup, wide bandwidth is not required. Also note that analogue phones have a maximim bandwidth need of 3K, hence even the low quality of G.723/5.3K, compared to the average analogue phone call, is superior
I think the data rates with overhead are as follow:
---------------------------------------------------------------- iLBC - 30 ms (a packet each 30 ms)
RTP payload: 50 bytes --> Rate: 13.3 Kbps RTP: 62 bytes UDP: 70 bytes IP: 90 bytes --> total rate: 24 Kbps
------------------------------------------------------------------- iLBC - 20 ms (a packet each 20 ms)
RTP payload: 38 bytes --> Rate: 15.2 Kbps RTP: 50 bytes UDP: 58 bytes IP: 78 bytes --> total rate: 31.2 Kbps
------------------------------------------------------------------- G.729 - 2 voice frame per packet
RTP payload: 20 bytes --> Rate: 8 Kbps RTP: 32 bytes UDP: 40 bytes IP: 60 bytes --> Total rate: 24 Kbps
------------------------------------------------------------------- G.729 - 4 voice frame per packet
RTP payload: 40 bytes --> Rate: 8 Kbps RTP: 52 bytes UDP: 60 bytes IP: 80 bytes --> Total rate: 16 Kbps
------------------------------------------------------------------- G.711 - 20 ms (a packet each 20 ms)
RTP payload: 160 bytes --> Rate: 64 Kbps RTP: 172 bytes UDP: 180 bytes IP: 200 bytes --> Total rate: 80 Kbps
El vie, 27-05-2005 a las 10:36, Iqbal escribió:
the 5.3, 6.3K are really theoretical, i dont think they include IP overheads, I used media_sessions.phtml, and looked at the actual calls per codec, and I dont think u can really get a good call without 50-70K, also most bandwidth providers (at least here in the UK) are asymmetrical, so even on a 128K, u could have problems.
Having said that I have done a nice call on xlite using ilbc on dial up.
Iqbal
Kofi Obiri-Yeboah wrote:
Just to add a few more details, Greger is right to point out the quality inferiority of G.723 compared to those of G711 and G729. In fact, in most VOIP deployments, in order to quarantee interoperability, a minimum bandwidth of 128K is specified. However to reach the wider "lower bandwidth areas" most service providers are opting for G.723 which uses either 5.3 or 6.3K. At this low bandwidth transmission needs, one could literally reach "dial up modem" equipped areas. in fact most VOIP phone hardware and software are begining to specify G.723 as their default codec. Note that until the direct media connection phase of a VOIP vall setup, wide bandwidth is not required. Also note that analogue phones have a maximim bandwidth need of 3K, hence even the low quality of G.723/5.3K, compared to the average analogue phone call, is superior
Serusers mailing list serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
which goes back to the point ur looking at a 50-70K as a rule of thumb...I just tell me users they need 128K to make it go nicely, on that they should be able to surf and talk at the same time
iqbal
Jorge Crichigno wrote:
I think the data rates with overhead are as follow:
iLBC - 30 ms (a packet each 30 ms)
RTP payload: 50 bytes --> Rate: 13.3 Kbps RTP: 62 bytes UDP: 70 bytes IP: 90 bytes --> total rate: 24 Kbps
iLBC - 20 ms (a packet each 20 ms)
RTP payload: 38 bytes --> Rate: 15.2 Kbps RTP: 50 bytes UDP: 58 bytes IP: 78 bytes --> total rate: 31.2 Kbps
G.729 - 2 voice frame per packet
RTP payload: 20 bytes --> Rate: 8 Kbps RTP: 32 bytes UDP: 40 bytes IP: 60 bytes --> Total rate: 24 Kbps
G.729 - 4 voice frame per packet
RTP payload: 40 bytes --> Rate: 8 Kbps RTP: 52 bytes UDP: 60 bytes IP: 80 bytes --> Total rate: 16 Kbps
G.711 - 20 ms (a packet each 20 ms)
RTP payload: 160 bytes --> Rate: 64 Kbps RTP: 172 bytes UDP: 180 bytes IP: 200 bytes --> Total rate: 80 Kbps
El vie, 27-05-2005 a las 10:36, Iqbal escribió:
the 5.3, 6.3K are really theoretical, i dont think they include IP overheads, I used media_sessions.phtml, and looked at the actual calls per codec, and I dont think u can really get a good call without 50-70K, also most bandwidth providers (at least here in the UK) are asymmetrical, so even on a 128K, u could have problems.
Having said that I have done a nice call on xlite using ilbc on dial up.
Iqbal
Kofi Obiri-Yeboah wrote:
Just to add a few more details, Greger is right to point out the quality inferiority of G.723 compared to those of G711 and G729. In fact, in most VOIP deployments, in order to quarantee interoperability, a minimum bandwidth of 128K is specified. However to reach the wider "lower bandwidth areas" most service providers are opting for G.723 which uses either 5.3 or 6.3K. At this low bandwidth transmission needs, one could literally reach "dial up modem" equipped areas. in fact most VOIP phone hardware and software are begining to specify G.723 as their default codec. Note that until the direct media connection phase of a VOIP vall setup, wide bandwidth is not required. Also note that analogue phones have a maximim bandwidth need of 3K, hence even the low quality of G.723/5.3K, compared to the average analogue phone call, is superior
Serusers mailing list serusers@lists.iptel.org http://lists.iptel.org/mailman/listinfo/serusers
.
In the following link there is a comparison between some codecs
http://www.speex.org/comparison.html
723: primary advantage is very low bandwidth requirements
Jorge Crichigno writes:
In the following link there is a comparison between some codecs
iLBC has also 20 ms mode which is not listed in the table.
-- juha
Yes, iLBC has the following variants:
------------------------------------------------------------------- iLBC - 30 ms (one packet each 30 ms)
RTP payload: 50 bytes --> Transmition rate: 13.3 Kbps RTP: 62 bytes UDP: 70 bytes IP: 90 bytes --> Transmition rate: 24 Kbps
------------------------------------------------------------------- iLBC - 20 ms (one packet each 20 ms)
RTP payload: 38 bytes --> Transmition rate: 15.2 Kbps RTP: 50 bytes UDP: 58 bytes IP: 78 bytes --> Transmition rate: 31.2 Kbps
-------------------------------------------------------------------
El vie, 27-05-2005 a las 09:45, Juha Heinanen escribió:
Jorge Crichigno writes:
In the following link there is a comparison between some codecs
iLBC has also 20 ms mode which is not listed in the table.
-- juha
-
Greger V. Teigre -
info＠beeplove.com -
Iqbal -
Jorge Crichigno -
Juha Heinanen -
Kofi Obiri-Yeboah