Re: [Megaco] A question about DTMF erasion

["Tom-PT Taylor" <[e-mail]>]

Gateway to gateway, the technical answer is RFC 2833, period.  RFC 2833, 
particularly as rewritten (draft-ietf-avt-rfc2833bis-12.txt, just passed 
WGLC) has advice on what to do with the audio signal.

In other applications you should view the SIP application framework:

draft-ietf-sipping-app-interaction-framework-05.txt



Kamitses, Jerry wrote:
> I just want to make sure that I'm on the same page regarding the 
> the discussion on when DTMF tones need to be removed from the audio 
> media at a MGW prior to encoding/packetizing the audio for transmission 
> in an RTP session
> 
> Consider the following situation where an IP network is used as a 
> backbone transport between separate PSTN networks and where the 
> RTP sessions are using a lossy compression codec to squeeze 
> the maximum performance from the IP NET.
> 			
> 	PTSN NET1/MGW1 <===== IP NET =====> MGW2/PSTN NET2
> 
> Two alternatives for getting DTMF tones to traverse the IP NET are
> (A) RFC2833 or (B) "manual" MGC detection/regeneration.
> 
> (A) When both MGW are RFC2833 capable then each MGW can detect DTMF tones 
> received on the PTSN network and encode them per RFC2833 as RTP packets for 
> sending to the alternate MGW.  It's my understanding that the DTMF tones in 
> this case must filtered from the PSTN media stream prior to applying the 
> lossy compression to avoid the possibility of duplicate tones being generated
> (since the RTP payloads used for the compressed voice and for the DTMF relay 
> could arrive at the far end MGW at different times).
> 
> 
> (B) If either MGW is not RFC2833 capable then the MGC at each MGW can 
> enable DTMF tone detection and then re-generate then via signals at the 
> far end MGW.  Is it the conclusion of the prior discussion that the 
> DTMF tones in this case would not be filtered from the audio media prior to 
> encoding and packetizing ?
> 
> 
> I can't quite tell from the prior discussion whether or not this specific 
> scenario was being addressed? 
> 
> 		
> 
> -----Original Message-----
> From: [e-mail] [mailto:[e-mail]]On Behalf
> Of Kevin Boyle
> Sent: Monday, December 05, 2005 2:38 PM
> To: [e-mail]; [e-mail]
> Subject: RE: [Megaco] A question about DTMF erasion
> 
> 
> RFC2833 is the industry standard for dealing with exactly the kind of
> scenario you are describing.  I think that if an implementation is going
> to support a compressive, lossy codec that it needs to implement this
> standard.
>  
> I am not sure why an external user would need to dial an external number
> on a PBX.  Why not just dial the number directly?  I am not sure I
> understand your scenario.  It seems to me that the only ones that would
> be wanted to access a PBX in the way described here are those who are
> served by the PBX.  In which case the PBX would be the originating end
> of the call, not the terminating end of the call.
>  
> If the MG alters the bearer stream, there are many potential
> consequences that are not being accounted for: what if there are other
> sounds on the line that are intended to pass end-to-end?  By removing
> the digits, the other data encoded with the stream is altered.  Further,
> there are many localities that require that media be unchanged, for the
> purposes of emergency services and lawful intercept.
>  
> I believe that the MG should not alter the bearer stream and the digits
> should pass through.
>  
> Kevin
> 
> 
> ________________________________
> 
> 	From: [e-mail] [mailto:[e-mail]]
> On Behalf Of [e-mail]
> 	Sent: Monday, December 05, 2005 7:20 AM
> 	To: [e-mail]
> 	Subject: RE: [Megaco] A question about DTMF erasion
> 	
> 	
> 
> 	Hi Kevin: 
> 	
> 	   If the the called is an PBX(connected to the called MG's
> analog line),and caller&called used G.723 to communication. Then caller
> may need to dail external phone number. The called PBX may can not
> detect the digit signal even though the digit signal is transmitted
> within the voice stream, because the signal is distorted when coded into
> G.723. But the called PBX still has the chance to detect the digit
> because sometimes the signal is not so distorted. 
> 	   So there are risks in transmite digit within the media
> stream. 
> 	   The responsibility of digit transmition falls on the MGC
> (forget RFC2833,suppose both MGs don't support RFC2833). The MGC may use
> packet 'dd' to request MG to detect digits and used packet 'tonegen' to
> request the called MG to play digit signal to the called PBX. 
> 	   From my view,  Once the MGC decide to collect numbers from
> MG, the MG SHOULD automatically erase the digit signal from the sending
> media stream to avoid repeated digits(one from media stream, another
> from signal). 
> 	    
> 	    
> 	
> 	Message: 2
> 	Date: Thu, 1 Dec 2005 04:23:01 -0500
> 	From: "Kevin Boyle" <[e-mail]>
> 	Subject: RE: [Megaco] A question about DTMF erasion
> 	To: <[e-mail]>, <[e-mail]>
> 	Message-ID: <34B3EAA5B3066A42914D28C5ECF5FEA40573F813@zrtphxm2>
> 	
> 	I have thought about this some, and my thoughts are that the MG
> would
> 	NOT remove the digits from the stream.  Think about what happens
> when
> 	the keypad is pressed while in an active call: The digit tones
> are
> 	carried all the way to the far end, even if there is something
> in the
> 	middle detecting the digits.  Therefore, I believe that the MG
> would not
> 	alter the media stream.  It would be incumbent upon the MGC to
> ensure
> 	that the digit collection is done at an appropriate place in the
> 	network, and to discard a repeated digit that is caused by dual
> 	collection.
> 	
> 	Kevin
> 	
> 	
> 	________________________________
> 	
> 	                From: [e-mail]
> 	[mailto:[e-mail]] 
> 	                Sent: Monday, November 28, 2005 11:50 PM
> 	                To: [e-mail]; Boyle, Kevin
> [NCRTP:3Z40:EXCH]
> 	                Subject: RE: [Megaco] A question about DTMF
> erasion
> 	                
> 	                
> 	
> 	                Hi Kevin: 
> 	                   I think You may have misunderstood my
> question. May be I have
> 	not clearified the 
> 	                situation. 
> 	                
> 	                   What I mean is : One MG has been requested to
> collect digits,
> 	and The MG detects 
> 	                a digit from the user's sending media stream.
> Then MG report the
> 	digit to MGC by 
> 	                signaling or hold the digit . But what confused
> me is whether
> 	the MG should 
> 	                automatically erase the DTMF signal from the
> sending media
> 	stream or Not. 
> 	                    
> 	                   If the MG do not erase the DTMF signal from
> the sending media
> 	stream, The called 
> 	                may receive repeated numbers: One from
> signaling, the other
> 	comes from The caller's 
> 	                Media stream. 
> 	                    
> 	                   I wonder if there are some rules defining how
> to handle this
> 	kind of situation. 
> 	                  
> 	                   Looking forward to your reply! 
> 	                        
> 	                   BEST REGARD! 
> 	                
> 	                   ShiCao 
> 	                Message: 3
> 	                Date: Sat, 26 Nov 2005 12:21:52 -0500
> 	                From: "Kevin Boyle" <[e-mail]>
> 	                Subject: RE: [Megaco] A question about DTMF
> erasion
> 	                To: <[e-mail]>,
> <[e-mail]>
> 	                Message-ID:
> <34B3EAA5B3066A42914D28C5ECF5FEA4055D11DB@zrtphxm2>
> 	                
> 	                Once a digit is reported to the MGC, I cannot
> think of a case
> 	where the
> 	                MG should transmit that digit again.  There are
> explicit
> 	statements in
> 	                the digit collection rules about when to hold
> digits and when to
> 	discard
> 	                them.
> 	                
> 	                Kevin
> 	                
> 	                
> 	                ________________________________
> 	                
> 	                                From: [e-mail]
> 	[mailto:[e-mail]]
> 	                On Behalf Of [e-mail]
> 	                                Sent: Thursday, November 24,
> 2005 7:16 AM
> 	                                To: [e-mail]
> 	                                Subject: [Megaco] A question
> about DTMF erasion
> 	                                
> 	                                
> 	                
> 	                                Hi: 
> 	                                
> 	                                  These days we encoute with
> such a problem: the
> 	caller recevied
> 	                repeated numbers due to the missunderstanding of
> DTMF erasion. 
> 	                                  
> 	                                  If Softswitch has requested MG
> to collect
> 	numbers , should MG
> 	                automatically enable DTMF erasion? 
> 	                                    
> 	                                  Is any packets defining
> enbable/disable MG's
> 	DTMF erasion? 
> 	                                  
> 
> 
> 
> _______________________________________________
> Megaco mailing list
> [e-mail]
> https://www1.ietf.org/mailman/listinfo/megaco
> 
> _______________________________________________
> Megaco mailing list
> [e-mail]
> https://www1.ietf.org/mailman/listinfo/megaco
> 
> 


_______________________________________________
Megaco mailing list
[e-mail]
https://www1.ietf.org/mailman/listinfo/megaco