[Asterisk-Users] dtmfmode=inband with G.729

[Steven Critchfield]

On Fri, 2004-04-02 at 10:12, Jim Rosenberg wrote:
> On Fri, Apr 02, 2004 at 08:52:09AM -0600, Eric Wieling wrote:
> > It's not asterisk, its the codecs.  Codecs other than ulaw and alaw will
> > distort continuous tones like DTMF.
> 
> Welllllllllll ...
> 
> At work we experience this with Cisco dial-peers over G.729: DTMF is
> erratic. But it's *NOT* inoperable. The way Asterisk does this, it
> doesn't even *try* to send the data through. I'd sure like that option,
> even if it might not register at the other end.
> 
> My Cisco-dial-peer-only connection users tell me that they often have to
> try a second time, but DTMF does usually work for them eventually.
> 
> "Might not resgister" does beat "Refuse to try" ...

No this is not better. If you allow the tones to be too relaxed you will
trigger them with certain peoples voices. This is known as talk off. We
as a group do not want talk off. In fact we should strive to avoid it as
it is considered a better to not have any talk off. 

> If you have an actual IVR application where errors matter, then of
> course you might decide you wouldn't want the risk of distored DTMF, but for
> simple things like picking an extension on a PBX where the consequence
> of an error is just a wrong number, why not give it a go?
> 
> Anyone who chooses dtmfmode=inband is knowingly choosing an option that
> is inherently error-prone.


-- 
Steven Critchfield  <critch at basesys.com>