Low audio accross VOIP

[ivan.marakovic]

Hi All,

We are using Cisco 3640 to send VOIP between two Avaya PABXs using G729r8 codec.

Some of the calls are very faint. Majority of calls are fine.

Using LLQ for QOS. Voip goes accross the 2Mbps HDLC.

Would input gain or output attenuation commands help here?

Also, show call active voice sometimes gives the following:

OutSignalLevel=0

InSignalLevel=0

ERLLevel=0

Has anyone seen this before?

Regards

Ivan

[kthorngr]

Hi Ivan,

Audio volume issues can be resolved with the input gain and output attenuation commands you mentioned. There is nothing you can do on the IPT side of the call to affect volume.

Kevin

[ivan.marakovic]

Thanks Kevin,

What people commonly use - input gain or output attenuation? Just to avoid causing an ECHO...

What do you think is the reson for the following output of "show call active voice" commnad?

OutSignalLevel=0

InSignalLevel=0

ERLLevel=0

Active call , but with signal levels at 0!?

Also, another one:

ACOMLevel=-8

OutSignalLevel=-61

InSignalLevel=-53

ERLLevel=0

Here I have ERLLevel at 0, but according to Cisco, you need this at least at 10 to 15dB!?

I appreciate your help.

Regards

Ivan

[carijit]

Hi Ivan,

ERL=0 is not a good sign as it leads to more echo.

Check the low Watermark and the High water mark on the sh call active voice traces.

Also try with input attenuation and output gain to bring up the ERL value.

If you are having a Call Manager setup then you can also play with

Audio Signal Adjustment into IP Network

Audio Signal Adjustment from IP Network

Echo TailLength (ms)

Minimum ERL (db)

values in the Call Manager Gateway Page.

Arijit

[ivan.marakovic]

Thanks Arijit,

Majority of calls will have ERL=20-50. Some of them have it at 0. Calls are going through multiple voice trunks and PSTN before they come to the router and some of them are very bad and with echo. What could be done to help this?

There is no Call Manager in our setup. We use 3640 gateways only, connected to Avaya PBX via E1 trunks.

The call I place from the phone attached to the local PBX has the following counts, and it sounds good.

HiWaterPlayoutDelay=70 ms

LoWaterPlayoutDelay=69 ms

ACOMLevel=-23

OutSignalLevel=-64

InSignalLevel=-41

InfoActivity=2

ERLLevel=30

What does the WaterPlayoutDelay mean?

The most frequent complaint is that the voice is vry faint, So to rectify this , I need to play with the input gain and attenuation. Which one is better to use in order not to get the echo worse?

And the last thing, I have Non-Linear-Processing (for echo cancelation) enabled on the voice port (by default).

Cisco says that NLP should be disabled in Double Talk calls. ( I think each call is a double talk call). So should I disable it?

syd-vgw1#sh voice port 2/0:15

ISDN 2/0:15 Slot is 2, Sub-unit is 0, Port is 15

Type of VoicePort is ISDN-VOICE

Operation State is DORMANT

Administrative State is UP

No Interface Down Failure

Description is not set

Noise Regeneration is enabled

Non Linear Processing is enabled

Non Linear Mute is disabled

Non Linear Threshold is -21 dB

Music On Hold Threshold is Set to -38 dBm

In Gain is Set to 0 dB

Out Attenuation is Set to 0 dB

Echo Cancellation is enabled

Echo Cancellation NLP mute is disabled

Echo Cancellation NLP threshold is -21 dB

Echo Cancel Coverage is set to 64 ms

Echo Cancel worst case ERL is set to 6 dB

Playout-delay Mode is set to adaptive

Playout-delay Nominal is set to 60 ms

Playout-delay Maximum is set to 250 ms

Playout-delay Minimum mode is set to default, value 40 ms

Playout-delay Fax is set to 300 ms

Connection Mode is normal

Connection Number is not set

Initial Time Out is set to 10 s

Interdigit Time Out is set to 4 s

Call Disconnect Time Out is set to 60 s

Ringing Time Out is set to 180 s

Wait Release Time Out is set to 30 s

Companding Type is A-law

Region Tone is set for AU

Station name None, Station number None

Translation profile (Incoming):

Translation profile (Outgoing):

Best Regards

Ivan

[carijit]

Hi Ivan,

Since your call is passing a number of trunks and PSTN troubleshooting can be a bit difficult.

HiWaterPlayoutDelay is a First-In, First-Out (FIFO) jitter buffer, high mark indicating the maximum depth to which the de-jitter buffer has adapted for this call.

LoWaterPlayoutDelay is a FIFO jitter buffer, low mark indicating the minimum depth to which the de-jitter buffer has adapted for this call.

Try setting the input gain -3 and output attenuation 3 in your voice ports.

Set the echo-cancel coverage to 32ms and enable echo cancel suppressor.

The NLP basically adds attenuation to the near-end signal after the echo is removed by the adaptive filter. So I dont think it will solve the problem by disabling NLP.

So try the above setting and get the sh call active voice trace and then disable NLP with the above setting and take the sh call active voice trace.

If possible post those traces.

Regards,

Arijit

[ivan.marakovic]

Thanks Arijit,

I have set the input attenuation of -3 on the voice-port 2/0:!5 on both cisco 3640 routers (both ends of voip call). I have not set the input gain yet. The audio volumes are better. I may have echo more noticable now.

I have got only one complaint about the faint voice so far, with the comment that this indial has always very faint audio. So, I think there is not much that could be done to rectify this.

Inpu gain -3 will attenuate the singnal? Is that right?

Why would you have input gain =-3 and output attenuation =3 on the same interface. Could this be done with only one of these two commands?

echo-cancel coverage is set to 64ms. Shall i go back to 32ms? I will not be able to cancel echos that are delayed more than 32ms.

echo suppressor command is not available for my voice port.

Does NLP performs echo suppresion as well?

Thank you very much for your help.

Best Regards

Ivan