| Summary: | ASTERISK-20773: fake DTMF or talkoff | ||||
| Reporter: | Marlon Costa (marloncos) | Labels: | |||
| Date Opened: | 2012-12-05 09:09:33.000-0600 | Date Closed: | |||
| Priority: | Major | Regression? | |||
| Status: | Open/New | Components: | Core/General | ||
| Versions: | Frequency of Occurrence | ||||
| Related Issues: |
| ||||
| Environment: | DAHDI Version: 2.6.1 Echo Canceller: HWEC, OSLEC Asterisk 1.8.18.0 i686 running Linux CentOS release 5.8 pci:0000:02:06.0 wct4xxp+ d161:0410 Wildcard TE410P (3rd Gen) | Attachments: | ( 0) beep_log_filtered.cap ( 1) beep_log_filtered.log ( 2) chan_dahdi_additional.conf ( 3) chan_dahdi.conf ( 4) dahdi-channels.conf ( 5) dtmf_rx.raw ( 6) dtmf_tx_rx.raw ( 7) init.conf ( 8) modules ( 9) system.conf (10) xpp.conf | ||
| Description: | Hi,
Many of our customers claims a beep when making or receiving calls. Our tech team enabled debug in our Asterisk and found that this beeps are DTMF tones. In asterisk logs it's possible see this things many times: {noformat} [Dec 5 12:40:13] DEBUG[5764] chan_dahdi.c: Begin DTMF digit: 0x41 'A' on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF begin 'A' received on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF begin passthrough 'A' on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DEBUG[5764] chan_dahdi.c: End DTMF digit: 0x41 'A' on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF end 'A' received on DAHDI/i1/21499100-fab, duration 51 ms [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF end accepted with begin 'A' on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF end 'A' detected to have actual duration 60 on the wire, emulation will be triggered on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF end 'A' has duration 60 but want minimum 80, emulating on DAHDI/i1/21499100-fab [Dec 5 12:40:13] DTMF[5764] channel.c: DTMF end emulation of 'A' queued on DAHDI/i1/21499100-fab {noformat} We done a wireshark capture to find out what is happening and it's possible see an RTP EVENT during these beeps see file attached. Could you please, help us to solve this issue? Thanks, Marlon | ||||
| Comments: | By: Richard Mudgett (rmudgett) 2012-12-06 12:27:16.669-0600 This sounds like a DTMF talk-off issue where talking triggers DTMF detection. Alec Davis recently made the DTMF detector options configurable in dsp.conf. See dsp.conf.sample for documentation. I think the version of Asterisk (v1.8.18) is new enough to have the ability added. By: Marlon Costa (marloncos) 2012-12-06 12:46:28.280-0600 Hi, Thanks for your answer, but I'm not sure that this is the issue. Here is the dsp.conf and we are still getting problem. [default] silencethreshold=256 dtmf_normal_twist=7.94 dtmf_reverse_twist=7.94 relax_dtmf_normal_twist=7.94 relax_dtmf_reverse_twist=7.94 dtmf_hits_to_begin=2 dtmf_misses_to_end=3 Thanks, Marlon By: Rusty Newton (rnewton) 2012-12-06 13:53:58.768-0600 Does this happen on extension to extension calls, or only on calls out through the TE410P and your service provider? Do your users "hear" a beep on the line? Or are you only seeing this indicated within Asterisk's debug? If this does not happen on extension to extension calls (only within Asterisk, as in SIP to SIP on local LAN), then you should contact your service provider to have them investigate. I've seen situations where a service provider had equipment sending out tones on the line on a regular frequency. By: Rusty Newton (rnewton) 2012-12-06 14:19:38.865-0600 Marlon, is this the same issue as reported in ASTERISK-20721 ? By: Alec Davis (alecdavis) 2012-12-06 14:21:27.398-0600 I'd change the the dtmf_hits_to_begin to 3 or 4. Each increment is 12.75ms. The aim here is to positively indentify a digit take a bit longer. This change may affect callerid recognition if DTMF is the method used in your country. Changing to 3, then means that the 'valid' period of a DTMF digit needs to be present for ~37ms. Changing to 4, ~51ms. For a period of time Asterisk tagged versions 1.8.10 to 1.8.16 had the hits=4 and misses=4. http://svnview.digium.com/svn/asterisk?view=revision&revision=349728 1.8.17 These were reverted back to the prior (before 1.8.10) settings of hits=2 and misses=3. 1.8.18 introduced the abilty to configure these settings. By: Marlon Costa (marloncos) 2012-12-07 05:18:18.731-0600 Hi all, We can hear this beep or DTMF sound, due this fact that we are collecting logs to investigate. This beep happens in all calls (incoming or outgoing) by passing on TE410P. All other calls (SIP to SIP or through other SIP interconnection carriers, this does not happen). This is a live Asterisk server and I'll change the configs said by Alec changing dtmf_hits_to_begin to 3 and 4 in our next off-peak period. Please, leve this ticket opened until. Thanks, Marlon By: Marlon Costa (marloncos) 2012-12-11 05:19:22.621-0600 Hi all, Thanks for your patience. I done the modifications that you suggested and the issue persists. Changed dtmf_hits_to_begin = 3 and 4 and the results was the same. [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF begin 'A' received on DAHDI/i2/33112864-127 [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF begin passthrough 'A' on DAHDI/i2/33112864-127 [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF end 'A' received on DAHDI/i2/33112864-127, duration 63 ms [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF end accepted with begin 'A' on DAHDI/i2/33112864-127 [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF end 'A' detected to have actual duration 60 on the wire, emulation will be triggered on DAHDI/i2/33112864-127 [Dec 11 09:05:04] DTMF[27471] channel.c: DTMF end 'A' has duration 60 but want minimum 80, emulating on DAHDI/i2/33112864-127 Could you please, help us on this? Thanks, Marlon By: Alec Davis (alecdavis) 2012-12-11 16:22:38.346-0600 Is it always the 'A' digit? How often during a call does it happen? How many times during a call? If it happens multiple time during a call, what is the period from the logs between beeps? By: Marlon Costa (marloncos) 2012-12-12 04:05:15.334-0600 Hi, No, it is possible see 'B', 'C' and 'D' digits also. This issue occour many times during a call and virtually in all calls. There is no specifc time to this issue, some times happens on start call, some times in the mittle and in other cases at the end. Thanks, Marlon By: Marlon Costa (marloncos) 2012-12-26 03:33:40.929-0600 Hi all, Any news about this TT? Thanks, Marlon By: Rusty Newton (rnewton) 2013-01-08 16:58:56.457-0600 Marlon, can you use [dahdi_monitor|http://docs.tzafrir.org.il/dahdi-tools/man/dahdi_monitor.8.html] to record the input (RX) audio stream before it gets to Asterisk? This will help out any developers who want to analyze the audio stream. How does your T1 connect to the TE410P? Is it coming straight from a service provider or is there another PBX in front of your Asterisk system (providing the T1) ? By: Marlon Costa (marloncos) 2013-01-11 06:39:15.943-0600 Rusty, see the files attached. About your queestion, our E1 is connected to a PSTN service provider. Thanks, Marlon By: Marlon Costa (marloncos) 2013-01-11 06:40:24.228-0600 Audio stream files | ||||