VoIP Mailing List Archives

[drew.ozier at gmail.com]

I've got a configuration where I receive inbound calls and dial out to a pre-determined 800-number based on the DNIS of the call. I set <action application="start_dtmf"/> and have everything set up so that DTMF only comes to me via inband. When I'm providing DTMF data to the IVR, it will recognize a single keypress as a double-tap. My FreeSWITCH logs only contain one DTMF entry, but when I listen to the receiving end of the call, I can hear a hiccup in the DTMF tone that is getting played. When I do not use 'start_dtmf', this problem goes away. I need inband DTMF detection, but I can't have it messing up the audio stream. Any thoughts?

-Drew Ozier

[testeador01 at gmail.com]

Check the freeswitch log when the dtmf are being received, most likely your device is sending them both as Dave says, maybe the telephone you're dialing with sends it both inband and rtp. What does the log says?

2009/5/26 Drew Ozier <drew.ozier@gmail.com (drew.ozier@gmail.com)>

[drew.ozier at gmail.com]

To clarify, I'm not running the IVR. I have a TDM T1 that comes in to an AudioCodes Mediant 1000 (SIP Gateway) which goes to a FreeSWITCH machine. I receive calls coming in off the T1 which goes through my Mediant 1000 which goes to my FreeSWITCH machine which makes an outbound call back through the Mediant 1000 to an IVR (bridging the inbound and outbound calls). I have (supposedly) configured the AudioCodes Mediant 1000 to pass the DTMF in-band, because I want to minimize the amount of modifications I make to the audio stream going through my system. I also want to be able to log when DTMF events occur for analysis purposes. If I do not have start_dtmf, then FreeSWITCH does not show me any DTMF in the logs, but when I have it on, it will.

Here's a DTMF event in my FreeSWITCH logs:
(I changed the debug message to a notice and added the callId and the nano time of the event)
2009-05-26 16:02:05 [NOTICE] switch_ivr_async.c:996 inband_dtmf_callback() 348323 61182225861200042042@10.1.10.10 (61182225861200042042@10.1.10.10) DTMF DETECTED: 5
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1322 do_2833() Send start packet for [5] ts=1217893935 dur=160/160/2000 seq=55199
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=320/320/2000 seq=55200
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=480/480/2000 seq=55201
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=640/640/2000 seq=55202
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=800/800/2000 seq=55203
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=960/960/2000 seq=55204
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1120/1120/2000 seq=55205
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1280/1280/2000 seq=55206
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1440/1440/2000 seq=55207
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1600/1600/2000 seq=55208
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1760/1760/2000 seq=55209
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send middle packet for [5] ts=1217893935 dur=1920/1920/2000 seq=55210
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send end packet for [5] ts=1217893935 dur=2080/2080/2000 seq=55211
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send end packet for [5] ts=1217893935 dur=2080/2080/2000 seq=55212
2009-05-26 16:02:05 [DEBUG] switch_rtp.c:1258 do_2833() Send end packet for [5] ts=1217893935 dur=2080/2080/2000 seq=55213

Does the fact that switch_rtp is sending packets mean that FreeSWITCH is sending out-of-band DTMF as well as recognizing the in-band?

Thanks for your quick replies, hopefully this will make my issue clearer.

-Drew

On Wed, May 27, 2009 at 10:22 AM, Milena <testeador01@gmail.com (testeador01@gmail.com)> wrote:

[brian at freeswitch.org]

You should try to always use out of band... what is the call path because it looks like a bridged call and the far end gets the inband and the 2833

/b

On May 27, 2009, at 10:02 AM, Drew Ozier wrote:

[drew.ozier at gmail.com]

Hey Brian,

It is a bridged call. Here's the majority of my dialplan:
<condition field="source" expression="mod_sofia">
        <action application="ring_ready"/>
        <action application="pre_answer"/>
        <action application="start_dtmf"/>
        <action application="set" data="effective_caller_id_number=${caller_id_number}"/>
        <action application="set" data="RECORD_STEREO=true"/>
        <action application="set" data="aleg_uuid=${uuid}"/>
        <action application="set" data="call_id=${strftime(%Y-%m-%d-%H-%M-%S)}_${destination_number}_${uuid}"/>
        <action application="set" data="outfile=$${base_dir}/recordings/${call_id}.wav"/>
        <action application="record_session" data="${outfile}"/>
        <action application="set" data="progress_timeout=20"/>
        <action application="set" data="call_timeout=90"/>
        <action application="set" data="bridge_pre_execute_bleg_app=start_dtmf"/>
        <action application="set" data="continue_on_fail=NORMAL_TEMPORARY_FAILURE"/>
        <action application="set" data="hangup_after_bridge=true"/>
        <action application="bridge" data="sofia/normal_customer/${outdial}@10.1.10.10 ([email]outdial%7D@10.1.10.10[/email])"/>
        <action application="bridge" data="sofia/normal_customer/${outdial}@10.1.10.10 ([email]outdial%7D@10.1.10.10[/email])"/>
</condition>

On Wed, May 27, 2009 at 11:12 AM, Brian West <brian@freeswitch.org (brian@freeswitch.org)> wrote:

[brian at freeswitch.org]

This is what starts the inband detector.. what is the far side?


/b



On May 27, 2009, at 10:25 AM, Drew Ozier wrote:

[brian at freeswitch.org]

I would configure both mediant 1000's to do rfc2833, or disable 2833 on your sofia profile and do 100% inband.. right now I have a feeling the mediant is getting the inband and the 2833 causing your problem.

/b

On May 27, 2009, at 10:45 AM, Drew Ozier wrote:

[testeador01 at gmail.com]

Are the configurations of the audiocodes set to send only in-audio dtmf? make sure everything is sending only in-audio

2009/5/27 Drew Ozier <drew.ozier@gmail.com (drew.ozier@gmail.com)>

[drew.ozier at gmail.com]

I'm sorry, far side meaning what exactly? The call flow is as follows:
TDM T1 -> AudioCodes Mediant 1000 -> FreeSWITCH -> AudioCodes Mediant 1000 -> TDM T1
FreeSWITCH gets an inbound SIP call and makes an outbound SIP call back to the Mediant 1000, bridging the two calls. The Mediant 1000 sends the call back out the T1 and that goes to my customer's IVR. I have no control over anything after it leaves my T1.
Is there some other piece of information I'm missing? Do you want my sip_profiles?

-Drew

On Wed, May 27, 2009 at 11:30 AM, Brian West <brian@freeswitch.org (brian@freeswitch.org)> wrote:

[brian at freeswitch.org]

I highly recommend you DO NOT use inband AT ALL if possible. out of band is king.

/b

On May 27, 2009, at 10:49 AM, Milena wrote:

[drew.ozier at gmail.com]

The AudioCodes machines are always a bit of a nuisance to get acting the way I want, but I think I have them sending all DTMF in-band. How do I disable 2833 in my sofia profile?

-Drew


On Wed, May 27, 2009 at 11:52 AM, Brian West <brian@freeswitch.org (brian@freeswitch.org)> wrote:

[drew.ozier at gmail.com]

Did a few TCP dumps, and before, I was seeing explicit DTMF packets coming out of my FreeSWITCH machine. Removing the only things I could find that had to do with 2833 in my internal and external sip_profiles (namely rfc2833-pt), I find that I don't see any mention of switch_rtp.c sending any DTMF packets, nor do I see any in the TCP dump.

This still hasn't solved my problem. I tested again, and with 'start_dtmf', I get an audible hiccup in the audio when DTMF occurs. When I don't use 'start_dtmf', the hiccup goes away.

Is there anything else that I should be turning off on FreeSWITCH to avoid this hiccup?

-Drew

On Wed, May 27, 2009 at 12:11 PM, Drew Ozier <drew.ozier@gmail.com (drew.ozier@gmail.com)> wrote: