[Asterisk-Users] VM low volume - testers needed

[Rich Adamson]

> On Monday 19 September 2005 12:38, Rich Adamson wrote:
> > The g(6) adds a 6 db gain for zap calls that end up recording a Voicemail
> > message.
> ...
> 
> > * 'g(#)' the specified amount of gain will be requested during message
> >          recording (units are whole-number decibels (dB))
> 
> How in the hell does that make any sense?  are your normal incoming calls 
> quiet too or just voicemail?  

Yes, see bug 2022 and 2023 for details, as well as 
  http://www.routers.com/asteriskprob/asterisk-config.htm
for a very detailed analysis of the problem.

I believe one of the more serious issues amounts to: if asterisk is
located a fair distance from the central office (-7db in my case), setting
the rxgain and/or txgain to any level that would be considered reasonable
for that loss (eg, rxgain=5, txgain=5), hugh amounts of echo result that
cannot be addressed through zapata.conf echo entris, and changing
compile options to agressive, etc, does not help. Its my believe 
(from working with several TDM users), the further one is from the CO,
the bigger the problem. (Or, short pstn cable lengths less then about
4 or 5db can almost always be addressed via parameters.)

The above workaround is very usable (assuming it works) when someone
calls in via the pstn and leaves a voicemail (which is already at
least 7db down plus their own pstn loss), and then I call in via the
pstn to retrive the voicemail (now 14db down PLUS the original callers
pstn loss), the audio is so faint its difficult to impossible to 
listen to. 

> > In my case, the asterisk box is located about 7db from the central
> > office. As noted in bug 2023 (and 2022), calls from an outside pstn
> > line coming into asterisk incure a 7db pstn loss (which can't be adjusted
> > for with rxgain and txgain as changing those values to something
> > reasonable generates echo).  Retrieving that VM message from an outside
> > location creates another 7db loss (now -14db down in total), making it
> > very difficult (if not impossible) to hear the message. (And, yes I've
> > gone through all the recommendations with wav vs gsm files, etc.)
> 
> I am not sure I understand why the txgain/rxgain isn't fixing it without 
> adding unacceptable echo...  this all seems very odd...  I mean for a test 
> you should be able to dial an echo() application and have extremely quiet 
> echoed audio... is this the case?

As an ex-telco transmission engineer, believe me I've done my homework
and some very solid testing with expensive well-calibrated test equipment.
As I've mentioned to Kevin, its almost like the TigerJet pci controller
on the TDM card is reversing bits six and seven (or something very odd
like that). Digium apparently now has a pci engineering type looking
at the issues, which I'm told is using a pci logic analyzer, etc.
 
> > The work around "only" kicks in if the call comes from a zap channel
> > and ends up in voicemail, adding a 6db gain to that recorded message.
> > No other channel types are impacted by this new parameter.
> 
> This is a HELL of a band-aid.

If you actually follow the logic that was originally stated in 2023,
this gain setting "is" highly useful for those systems that are 
further away from the CO (as mentioned above). For those closer to
the CO, it has zero value.

Rich