[asterisk-dev] [svn-commits] kpfleming: branch 1.4 r3490 - /branches/1.4/

[Steve Underwood]

Kevin P. Fleming wrote:
> Steve Underwood wrote:
>
>   
>> Nice, but a nation by nation twist is slightly questionable. Applying 
>> twist at the source only seems to be appropriate for analogue paths. The 
>> way some countries specify it is kinda vague. A number of DTMF 
>> generators have a 2dB twist locked into them, but they were only ever 
>> intended for analogue connections. I would question whether a 2dB twist 
>> should ever be applied over a digital path.
>>     
>
> So what happens when that digital path becomes analog at the far end?
> I'd be quite surprised if the far-end channel bank modified the audio to
> achieve this.
>   
The -2dB twist isn't intended to achieve a -2dB twist at the far end. 
Its to avoid it going highly positive, with the inevitable high 
frequency rolloff you get on analogue lines. Its just a compromise 
fudge, not something deeply engineered. If your local end isn't 
analogue, it makes no sense, regardless of what lies beyond.
> In any case, I've only modified the zonedata for Brazil, because we've
> gotten specific instructions from the regulatory agency there that there
> must be 2dB twist whenever we generate DTMF. The rest of the zones are
> still at -10/-10 (as they were before) and we'll only modify them when
> we're really, really sure we should :-)
Its very common practice all over the place. As I said before, some DTMF 
generators cannot even turn off that -2dB twist for any locale. You 
might find using that twist in any locale where you use analogue 
connections improves your DTMF reliability. The Brazilian spec is 
obviously based on a phone on an analogue line generating the DTMF. Have 
you been testing there with analogue or digital connections?

Regards,
Steve