[VoIP] DP-compatible ATAs

[ikjtel]

In Message-ID 45945B97.7050609 at jps.net, Steph Kerman
wrote:

> [snip]  However the CNET calls I've made so far
through Greg's 212 PSTN portal don't sound all that
wonderful to me.

The question of how to improve CNET (from an
audiophile perspective) is an interesting one, it
should first be noted that a PSTN portal may not be
the most fair way to judge.  Would want to know more
details about it (whether any additional A/D
conversions occur, whether any compressing codecs are
in use, how it addresses the very-tricky echo issues,
etc).

The way a CNET office sounds to well-connected callers
from the outside world is mainly determined by the FXO
card in use.  The X100P (and clone) FXO's are all crap
IMHO - they're optimized for use as data modems, not
voice; their audio dynamic range is known to be
limited; they cut out on DC and inductive spikes, etc.
 I don't have the hardware to test but would fully
expect a channel bank FXO port to outsparkle an X100P,
audio-wise.

> [snip] If there is no 2W segment in the call because
it is 4W-IP end-end, echo should be non-existent. 
Echo results from the 2W-4W junctions.

This didn't make sense to me.  There's *always* at
least one 2W segment in this case (at the ATA).

In Message-ID 45944B66.7020809 at jps.net, Steph Kerman
wrote:

> [snip]  I'd prefer to have the DP outpulsing of real
wirespring sender audible to the caller, not ersatz
"recorded" outpulsing sounds from an * box.

Agreed.  However, easier said than done.  See below...

In Message-ID 45954C42.6563.599D269 at localhost Chad
Perkins wrote:

> I can understand that, however Asterisk nativity
will not add nor remove outpulsing "sounds"; whatever
goes in one end comes out the other ...

I do not mean to pick on Chad, but that isn't correct.
 Both zaptel and asterisk have code specifically to
suppress outpulsed DP (rotary) dial digits.  A while
back I released a patch to defeat this, in the
interests of those demented souls (myself included)
who wanted to hear the real dial tone and outpulsing. 
Last time I checked there was at least one CNET office
(try 366-3514) that still had this in effect.  The big
letdown was that the SxS selector "clunk" sound (the
"D" relay operating) was suppressed (by the X100P
hardware) although the selector level hunting sounds
were audible ...

> [snip] The VoIP part of CNET is a TCP/IP medium and
I would suggest is neither 2W or 4W per se

Conceptually the VOIP portion is treated as 4W, the
two directions are processed separately...

Max

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com