I've deployed 7 Asterisk based PBXs now for small businesses and am getting pretty comfortable with getting a reliable setup. Now I'm trying to dig deeper and start tweaking things.
These are entirely SIP based using a mix of SIP providers for trunks. We're using Polycom ip560s and ip650s for endpoints. I'm using Piaf 1.7.9 with Asterisk 1.4.38. We're using G711u for codecs everywhere.
1) Is echo cancellation needed at all in this case? It looks to me like it should be the ITSP providing the trunk doing the echo cancellation correct? If echo cancellation is useful is it better on the endpoints, Asterisk, or both?
2) I'm trying to reduce a slight delay one of my clients noticed. It's nothing major, but if it can be tweaked a little I'd like to learn. The Polycoms have a default minimum jitter buffer of 40ms. I assume this jitter buffer is used only for traffic between the PBX and the phone. The internal network should have essentially no jitter, especially not 40ms of jitter. Can I reduce this to 0ms on the Polycoms? How does Asterisk's jitter buffer come in to play? It looks like it's off by default.
3) I'd found cases with people using Linksys ATAs saying that changing the RTP packet size to 10ms gave them better call quality. I haven't found much about changing the packet size in Asterisk, can it be done for the endpoints and the trunks? Are there any issues to consider? We have more than enough bandwidth so that's not a concern.
4) I've noticed a little background static on the phone that goes away when people are talking. I assume this is the silence suppression kicking in. Is this generated by Asterisk or the Polycoms and if it's Asterisk and is there a way to reduce the noise a little? The first site I tested with I used Trixbox and ip650s and didn't notice the static so I've always suspected an Asterisk setting.
5) Post up any other tips and tricks that might improve audio quality!
I did find running dahdi_test resulted in an average of about 9.96x%, I've ordered a Sangoma UT50 to see if that raises the score any. Most results were 99.997 to 99.998 with occasional dips to 99.6xx which killed the average.
These are entirely SIP based using a mix of SIP providers for trunks. We're using Polycom ip560s and ip650s for endpoints. I'm using Piaf 1.7.9 with Asterisk 1.4.38. We're using G711u for codecs everywhere.
1) Is echo cancellation needed at all in this case? It looks to me like it should be the ITSP providing the trunk doing the echo cancellation correct? If echo cancellation is useful is it better on the endpoints, Asterisk, or both?
2) I'm trying to reduce a slight delay one of my clients noticed. It's nothing major, but if it can be tweaked a little I'd like to learn. The Polycoms have a default minimum jitter buffer of 40ms. I assume this jitter buffer is used only for traffic between the PBX and the phone. The internal network should have essentially no jitter, especially not 40ms of jitter. Can I reduce this to 0ms on the Polycoms? How does Asterisk's jitter buffer come in to play? It looks like it's off by default.
3) I'd found cases with people using Linksys ATAs saying that changing the RTP packet size to 10ms gave them better call quality. I haven't found much about changing the packet size in Asterisk, can it be done for the endpoints and the trunks? Are there any issues to consider? We have more than enough bandwidth so that's not a concern.
4) I've noticed a little background static on the phone that goes away when people are talking. I assume this is the silence suppression kicking in. Is this generated by Asterisk or the Polycoms and if it's Asterisk and is there a way to reduce the noise a little? The first site I tested with I used Trixbox and ip650s and didn't notice the static so I've always suspected an Asterisk setting.
5) Post up any other tips and tricks that might improve audio quality!
I did find running dahdi_test resulted in an average of about 9.96x%, I've ordered a Sangoma UT50 to see if that raises the score any. Most results were 99.997 to 99.998 with occasional dips to 99.6xx which killed the average.
Phone System 