I have a customer that's escalating an issue with Cisco. TAC is working on it. Here's the scoop...
It's best to begin with an excerpt from a Cisco tech doc.:
"The Cisco IP Phone 7960 supports comfort noise. When Voice Activity Detection (VAD) is enabled, the Cisco IP Phone 7960 generates a low level white noise hiss to simulate the background circuit noise users experience on non-IP connections. Generated comfort noise is only heard on the handset of the station generating the noise; it is not transmitted to the station at the other end of the call. If VAD is disabled, the Cisco IP Phone 7960 provides a continuously streaming, end-to-end connection. VAD is enabled by default.
The comfort noise level is determined either by information contained in Silence Information Description (SID) frames or by calculations performed on previous information packets. If a call is made from one Cisco IP Phone 7960 to another Cisco IP Phone 7960, the transmitting device sends SID frames that enable the comfort noise generator on the receiving device to generate the proper level of comfort noise. If a call is made between a Cisco IP Phone 7960 and a device that does not send SID frames, the Cisco IP Phone 7960 calculates and generates the proper comfort noise level based on the background noise of previous information packets."
Our customer has a Call Manager, Cat 4000 AGM w/ MFT-T1 configured for PRI voice. VOIP dial peers have VAD disabled, comfort noise is disable on the voice ports, and Call Manager's SilenceSuppressionSystemWide is off (set to "F").
Symptoms are the following:
IP Phone to IP Phone - no hiss (works as it should)
IP Phone to PSTN (inbound or outbound call during dead silence) has a low-level hiss heard on the IP Phone.
Previously suggested solutions:
Disable VAD - done, no help
Disable comfort noise - done, no help
Upgrade the IOS - done, no help
Disable Silence Suppression - done, no help
Change the input gain/output attenuation - done (great for echo, doesn't do squat for hiss)
Disable/Enable SilenceSuppressionSystemWide - done, no help
I have done quite a bit of research on Cisco's site. There were some tech docs related to bugs on gateways where even though VAD is off, MGCP or H323 turns it on in error. Other than that I have nothing to go on.
Incidentally, I have a 7835 and 7750 in our own lab. I have the same background hiss on all gateway calls. It's low, you can barely hear it, and it's only during silence, but the customer says it's unacceptable (and finds great joy in pointing out that their old PBX doesn't do it).
Any help would be appreciated,
Sr. System Engineer
CCNA, CCDA, CIPT
Have a simillar setup here (a CM & a PBX). IP phone to IP phone is just fine (i.e. no hiss), IP Phone to PBX extension= no hiss, BUT..... IP phone to PSTN = low hiss, PBX to PSTN= low hiss... Seems to me that the hiss is originating from the PSTN and the IP phone & PBX is doing everything it is supposed to in re-playing it. I'll bet it did that too on the customers OLD PBX, he just dosen't want to admit it.
I know it's not coming from the PSTN from a simple test. In our lab here I unplug a pots line from the gateway, plug it into an analog phone and viola - no hiss.
I'm pretty sure the IP phone is generating it. Telcos intentionally put some line noise over the lines. We all know that when you use cell phones and here dead silence you end up asking "are you still there?". I assume the deliberate hiss in this case is to let you know the line is active. However, the hiss from the IP phones is far greater than anything I hear coming from the PSTN.
Incidentally, I also tried a call through our Nortel PBX and during dead silence, it's clear. You can hear a very very low hiss - but nothing like what is coming from the IP Phone.
I'll bet you get the hiss if you make a call from an IP phone to the PSTN, but not from the PSTN to the same IP phone, right? In that case......
I assume you have dial-peers on the Cat4k AGM something like this:
dial-peer voice 4000 voip
ip precedence 5
session target ipv4:a.b.c.d
(or something like that). And a pots dial-peer like this:
dial-peer voice 9 pots
(or something like that). In this case I will assume that for ease of AGM dial-peer configuration that you have setup a 9.@ Route Pattern on CM that discards no digits when the setup is sent to the AGM, so you can simply use a destination-pattern of 9 and be done with it.
What I suggest you do is on one of the voip dial-peers you have that points to CallManager and has no vad already configured, that you add the line incoming called-number 9. This will force incoming VoIP calls to the AGM from the CallManager to disable VAD.
What you can do to verify whether or not VAD is enabled for a call is to do a show call active voice on the AGM while you have a call from an IP phone to the PSTN up. It will tell you whether or not it's enabled (near the end of the output). You may want to look at that before and after you use the command I suggested.
Thanks for the reply...
Here's the AGM's dial-peer & voice port config
input gain 4
output attenuation 4
echo-cancel coverage 24
dial-peer voice 200 pots
dial-peer voice 400 voip
progress_ind setup enable 3
session target ipv4:10.100.10.1
The route pattern on the CM is 9.@ w/ PreAt stripped. I have another route pattern for 911 to use local pots lines.
I just called to customer to check the inbound theory. They say the hiss is still there. With that in mind, should I still attempt your config changes above? Also, are my 9.@/911 CM route patterns config "compatible" with your incoming called-number 9 command?
There is one other service setting for CallMananger you should try turning off: SilenceSupressionWithGateways
I did notice you turned it off system wide. I always turn off both and never have a hiss problem with a VG200/2600/3600/6608 gateway.
The docs on this setting say it is not applicable for H323 or MGCP gateways. Go figure. Anyhow, I've already tried this before with no success.
p.s., TAC still has this case and has told me they're escalating it to the product team. We'll see. Any other ideas out there?
There would be a couple config changes required with respect to the dialplan on the AGM and the CM, but just to find out if my idea is on the right track, can you please have the customer get an active call where the hiss is heard, and while they are on the phone, do a show call active voice and look to see if VAD is enabled or disabled? You would want it to be disabled at all times, but I believe without doing some kind of config with incoming called-number x then it can still happen.
Got the same problem.
Did what Dave told you (a fake dial-peer with VAD and Incoming called-number).
The sound is quite all right, but not as good as a Legacy PBX phone.
Did you get some news from TAC ?
We had the same problem, but also could not get rid of echo (heard only on ip phone side). Fix it the following way: Added VOIP dial peers on each gateway that basically "mapped" (for lack of knowing a better technical word) to each internal IP phone extension, for example 1.. and 21.., set it not to use VAD. The result removed all injected noise and echo. We added each one twice, to ensure it would remain during failover (each pointed to one of the CMs) Hope this helps point you in the right direction.
A TAC engineer resolved a similar problem for us today. By adding incoming called-number . to the dial-peer voip, the hissing is no longer there.
dial-peer voice 402 voip
incoming called-number .
I believe that fix is only useful when utilizing h323 gateways as well as any other ios gateway/inbound trunk device. The reason being in a h323 gateway there is a default dial peer that is not listed that has VAD enabled. It makes a default match with any dialed number that is not matched with a listed dial peer The incoming called number. line makes sure that all calls match at least one peer.