asterisk users - Feb 2004 - iax2 jitter stats confusion

I have been kind of tracking IAX2 calls and trying to measure performance 
with a given "iax2 set jitter" command.  My default is 250ms..

When a call is in progress I'll be watching it at the console with
"iax2
show channels"  Here are my stats from one particular call:

66.225.202.72    benshaw     00001/16413  00048/00035  00489ms  0221ms  ILBC
66.225.202.72    benshaw     00001/16413  00137/00125  00487ms  0270ms  ILBC
66.225.202.72    benshaw     00001/16413  00137/00125  00491ms  0269ms  ILBC
66.225.202.72    benshaw     00001/16413  00141/00129  00487ms  0241ms  ILBC
66.225.202.72    benshaw     00001/16413  00141/00129  00480ms  0235ms  ILBC
66.225.202.72    benshaw     00001/16413  00143/00131  00480ms  0256ms  ILBC
66.225.202.72    benshaw     00001/16413  00144/00132  00492ms  0268ms  ILBC
66.225.202.72    benshaw     00001/16413  00152/00140  00487ms  0472ms  ILBC
66.225.202.72    benshaw     00001/16413  00154/00142  00507ms  0473ms  ILBC

Now I figured the guy would be coming up to my office shooting but when I 
asked him how the call was he said "perfect." -- now he knows he's
on a
VOIP call but he had no idea of the jitter and lag here... 

So I suppose my question is "huh?"

How can I have such poor jitter and yet have this guy (not a techie) claim 
the call was perfect?  Neither he nor the guy on the other end (PSTN 
through NuFone) had any issues about the quality.

I don't want to look a gift horse in the mouth, so to speak, but I would 
like to know how to measure call quality; I thought jitter was a pretty 
good indicator.

Regards,
Andrew

> When a call is in progress I'll be watching it at the console with
"iax2
> show channels"  Here are my stats from one particular call:
> 
> 66.225.202.72    benshaw     00001/16413  00048/00035  00489ms  0221ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00137/00125  00487ms  0270ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00137/00125  00491ms  0269ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00141/00129  00487ms  0241ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00141/00129  00480ms  0235ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00143/00131  00480ms  0256ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00144/00132  00492ms  0268ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00152/00140  00487ms  0472ms 
ILBC
> 66.225.202.72    benshaw     00001/16413  00154/00142  00507ms  0473ms 
ILBC
> 
> Now I figured the guy would be coming up to my office shooting but when I 
> asked him how the call was he said "perfect." -- now he knows
he's on a
> VOIP call but he had no idea of the jitter and lag here... 
> 
> So I suppose my question is "huh?"
> 
> How can I have such poor jitter and yet have this guy (not a techie) claim 
> the call was perfect?  Neither he nor the guy on the other end (PSTN 
> through NuFone) had any issues about the quality.
I'll take a stab at this, but you'll probably get as many opinions as
there
are readers.

Jitter is reflective of the variation in packet delay between end points,
not a measurement of audio quality. If none/few of the packets are dropped,
the user wouldn't even notice other then maybe a click or something. The
fact that delay exits and the variation in the delay is rather large doesn't
mean it "will" impact quality. However, the opposite might be true: if
quality
were poor and you found jitter to be very high, then jitter is likely the
"symptom" and not the root cause.

If your user would have an analog pstn call going on simultanously, he
"would" notice the significant delay. Likewise, if the VoIP call had
any
echo characteristics, he'd notice the delay.

It would appear the necessary data packets are arriving in such a way as to 
allow the jitter buffer to do what its supposed to do, and apparently doing 
it very well. Unless you're using satellite, it would appear the delay
numbers
are rather high in my opinion.

Rich