[Asterisk-Dev] zaptel "rest value" of 0xff is not appropriatewhen ALAW is in use

[steve at daviesfam.org]

On Mon, 12 Sep 2005, Michael Procter wrote:

> steve at daviesfam.org wrote:
> >
> > Hi Kevin, others,
> > 
> > See http://bugs.digium.com/view.php?id=5186 for the fix for this issue.
> >
> > Steve
> 
> I agree with your analysis, but from memory, 0x55 is the traditional idle
> pattern for A-law, not 0xD5.
> 
> It has been a few years since I worked at that level in the stack, but
> surely there is someone else on the list who can confirm or deny the
> preferred idle pattern!

But, you see, the span isn't necessarily idle.

It might be.  Or, it might be that Asterisk just isn't writing audio to 
that particular channel right now (like, for instance, its doing Wait(1)).

Right now what you see is that, DURING AN ACTIVE CALL, every time Asterisk 
stops sending audio on the channel for any reason, the level jumps to a 
non-zero value.

Anywa - I see that, whilst AST_ALAW(0) gives you 0xD5, that the actual
reverse conversion of 0xD5 is +8.  The reverse value of 0x55 is -8.  So
either is as close as it gets to 0.

In the ALAW case, 0xFF represents 0 exactly.

So if 0x55 is the traditional value for an idle span on alaw, then we use 
use 0x55 in preference to 0xD5.

I take it that you do agree, though, that 0xFF is not the right value to 
use on an ALAW channel?

Steve