- This topic is empty.
9th February 2001 at 16:12 #19941JoeGuest
When designing VoIP hardware, which CODECS should be supported?
I know that G.711 has the best quality but don’t how this balances against G.729 which maximizes number of SIP sessions that can supported in the “pipe”.
Along with that are royalty payments for G.729. What about G.726? Would appreciate any info on this. Thanks.12th February 2001 at 23:57 #19942RGuest
Cover your bases:
G.723.1 most widely used for PCs and holds a MOS score of ~3.3
G.729 most widely used for compressed voice between gws with a MOS of 3.9
G.711 MOS 4.1 which is considered toll quality.
G.726 good to support but has no real benefit as the ADPCM algorithm is not as robust as the CELP flavour.
If I was building, G.711 and G.729 would be a must at release. Might as well go with G.729a as you can squeeze more calls out of the DSPs due to the complexity of the codec.
G.729 maxes the number of RTP sessions not SIP sessions.15th February 2001 at 06:37 #19943myordGuest
As always it’s a question of compromise between Bandwidth and Quality.:-)
Depends on what is the most significant issue in your case.
I would go the same way as R ..one waveform codec (G.711) and one source codec (G.729).Especially the last one with compressed RTP (cRTP)and VAD will be cool solution for bandwidth and quality.
Any suggestions for G.723.1 ??28th February 2001 at 18:13 #19944RGuest
I personally dont like the G.723.1 codec as its quality is poor.
cRTP is a cisco proprietary mechanism so I doubt you’ll be able to implement that but VAD or silence supression is a good choice.1st March 2001 at 18:57 #19945myordGuest
Yep..I am on the same opinion.
I don’t like G.723.1 too because of the quality..and i was interested to see if it’s only my assumption. cRTP is mechanism implemented in Cisco solutions for VoIP..and i consider it as not of a bad approach to the problem.
I am interested in some equipment /vendors/ for testing and emulating the RTP stream..and the IP network./different from Empirix../ Thanks!!