[pjsip] Fw: Sound issues: strange samplerates?

Alain Totouom alain.totouom at gmx.de
Thu Apr 7 06:00:23 EDT 2016


Hi Oliver,

the callee (Fr!tzBox) did alter the supported codec list (offer/answer) 
between his provisional (183/INVITE/cseq=6911) and his final response 
(200/INVITE/cseq=6911) and did decline the callers request 
(UPDATE/cseq=6912) to use the previously negotiated one (G722).
A more robust solution in a case like this, would i.E. imply caching the 
answer-codec-list and updating the offer-codec-list for subsequent calls 
to the same endpoint…


00:58:12.143   pjsua_core.c  .......TX 1276 bytes Request msg 
INVITE/cseq=6911 (tdta003BBA90) to UDP 192.168.1.1:5060:
[..]
m=audio 4000 RTP/AVP 98 97 99 104 3 0 8 9 96
[..]

00:58:12.347   pjsua_core.c  .RX 755 bytes Response msg 
183/INVITE/cseq=6911 (rdata003392C4) from UDP 192.168.1.1:5060:
[..]
m=audio 7086 RTP/AVP 9 0 8 96
[..]

00:58:12.347  pjsua_media.c  ......Audio updated, stream #0: G722 (sendrecv)
00:58:34.052   pjsua_core.c  .RX 1001 bytes Response msg 
200/INVITE/cseq=6911 (rdata003392C4) from UDP 192.168.1.1:5060:
[..]
m=audio 7086 RTP/AVP 0 8 96
[..]

00:58:34.104   pjsua_core.c  ....TX 815 bytes Request msg 
UPDATE/cseq=6912 (tdta003D31C8) to UDP 192.168.1.1:5060:
[..]
m=audio 4000 RTP/AVP 9 96
[..]

00:58:34.155   pjsua_core.c !.RX 353 bytes Response msg 
488/UPDATE/cseq=6912 (rdata003392C4) from UDP 192.168.1.1:5060:

Cheers,
Alain

On 07/04/16 11:14, Andreas Ahland wrote:
> HI Oliver,
>
> I remember we had issues with the FritzBox as well as ours did not send the media type which was negotiated in advance. This was clearly a fault in the FritxBox. We got our setup working by using G711a only, i.E. disable G722 and G711u.
>
> Mit freundlichen Grüßen
>
> Dr.-Ing.
> Andreas Ahland
> CTO
> Technischer Leiter
> Telefon : +49 251 6183-196
> Telefax : +49 251 6183-197
>
>
>
>
> Von: pjsip [mailto:pjsip-bounces at lists.pjsip.org] Im Auftrag von Oli Kah
> Gesendet: Donnerstag, 7. April 2016 01:13
> An: pjsip at lists.pjsip.org
> Betreff: Re: [pjsip] Fw: Sound issues: strange samplerates?
>
> Hi Bill,
>
> thanks for answering so promptly =)
> I have created the log and yes, there seem to be lots of suspicious lines like "strm003CAB04  Bad RTP pt 0 (expecting 9)" but the rest of the log does not tell me that much and looks mostly as expected?!
>
> The three major problems currently for me are:
> - it takes much too long to really establish a call (10 seconds are not normal)
> - the sound received by the callee sounds strange, as if the sample rate is wrong
> - the callee is NOT heard at all on the caller's side!
>
> Do you see what issues are there in the log that might explain this behavior? And what needs to be done to fix that?
>
> Thank you so much!
>
> Cheers,
> Oliver
>
> Gesendet: Mittwoch, 06. April 2016 um 22:52 Uhr
> Von: "Bill Gardner" <billg at wavearts.com<mailto:billg at wavearts.com>>
> An: "pjsip list" <pjsip at lists.pjsip.org<mailto:pjsip at lists.pjsip.org>>
> Betreff: Re: [pjsip] Fw: Sound issues: strange samplerates?
> Hi Oliver,
>
> Please generate a pjsip logfile (level 4 should suffice) and send, there may be clues in there.
>
> Bill
>
> On 4/6/2016 4:26 PM, Oli Kah wrote:
> Hi Bill, hi everyone,
>
> thanks for answering but I can confirm that the config_site.h settings are NOT the problem. I have now recompiled the lib with nothing in config_site.h.
> The result is exactly the same.
>
> To illustrate what the called person hears I have attached an mp4 which also shows the progress of the Python app in form of PyCharms debug log.
>
> There are important issues that can be seen + heard. Note that I am saying "1-2-3" from the beginning(!) of the video until the very end without ever stopping. Only the receiving side can be heard voice-wise (or not as you will see/hear).
>
> 1) When picking up the phone the sound CANNOT be heard on the receiving side although I am continously saying "1-2-3". It takes roughly 10(!!) seconds before the other side can hear me at all!!
> 2) It also takes 10 seconds after pickup before the call confirmation phase is reached. This is extremly slow and totally unexpected.
> 3) When the other side finally can hear me I sound like a monster... the sample rate seems to be off - I can't otherwise explain the strange sounding voice!
>
> The Python code of this mini app is attached to this email once more.
>
> I don't think I am doing anything exotic. It does not work as expected though. If I do the same with an app like Phoner (see http://www.phoner.de/download_en.htm) none of these problems occur within the same environment and using the same accounts and phone numbers. So it has to do with pjsip lib somehow.
>
> Any ideas?
>
> Thank you.
>
> Cheers,
> Oliver
>
> Gesendet: Montag, 04. April 2016 um 20:49 Uhr
> Von: "Bill Gardner" <billg at wavearts.com><billg at wavearts.com>
> An: pjsip at lists.pjsip.org
> Betreff: Re: [pjsip] Fw: Sound issues: strange samplerates?
> Hi Oliver,
>
> I think you should try a completely default configuration, i.e. use an empty config_site.h file. Your config_site.h params may be causing problems.
>
> Regards,
>
> Bill
>
> On 4/4/2016 2:37 PM, Oli Kah wrote:
> Hmmm, no one?
>
> Is there some sort of forum somewhere where to post things like these??
>
> Thank you :)
>
> Cheers,
> Oli
>
> Gesendet: Freitag, 01. April 2016 um 21:38 Uhr
> Von: "Oli Kah" <mj_fn at web.de><mj_fn at web.de>
> An: pjsip at lists.pjsip.org<mailto:pjsip at lists.pjsip.org>
> Betreff: Sound issues: strange samplerates?
> Hi there,
>
>
> I am new to this list and want to say "Hello" to everyone listening :)
>
> My issue using pjsib is rather strange. When calling someone (I tested it using two of my own telephone numbers) I get normal audio first during early call stages and then when the call is finally confirmed the audio rate suddenly is half or so. The voice then sounds monsterish and the remote site is no longer audible via speakers. This happens every time - reproducible!
>
> What I did:
>
> I have successfully compiled pjsua (release build) for Python using Visual Studio 2015 Community linking pjsua to Python 2.6 lib, 32bit running everything on Windows 8.1 Pro 64bit. So far so good.
> These values were used for my config_site.h:
>
> #define PJMEDIA_CONF_USE_SWITCH_BOARD 1
> #define PJMEDIA_SOUND_BUFFER_COUNT 8
> #define PJMEDIA_SND_DEFAULT_REC_LATENCY 50
> #define PJMEDIA_SND_DEFAULT_PLAY_LATENCY 50
>
> The attached python code shall define a simple SIPPhone that will be used in my larger scale application. It's unfinished but I ran into these unresolvable audio problems and hope that you might help me. In its current form the SIPPhone is totally useless!
>
> During testing I have been using the onboard audio card of my mainboard with some Bose speakers and a Samson UB-1 USB microphone which is connected to the Windows-PC using an USB 2.0 port. For the "remote" side I use a dedicated VOIP telephone from Grandstream.
>
> The test code in my attached file is then run using Pycharm + Python 2.6 runtimes. For testing you must replace the numbers + credientials by valid values on your side. Also note that you might have to specify another domain (mine here is "fritz.box").
>
> The called Grandstream rings. Taking up the phone then starts the call confirmation which is going unexpectedly slow. It takes quite some time (between roughly 3-10 seconds) before the "confirmed" status is reached although the audio starts working before that. The scenario is running on my LAN where the SIP server (a FritzBox) is also running here. So quite strange why this takes so long?
> Right after phone pickup the audio can't be heard at all (me constantly talking after phone pickup!) - in both directions! Then after a short time (1-4 sec) the voice can be heard normally and understandably as expected in both directions. But this only works until the "confirmed" status is reached for the call. When it is reached there is no longer any sound coming from the PC speaker and the voice heard on the Grandstream is monsterish (half sample rate?!). The bidirectional audio is lost and the remaining audio is really bad.
>
> The observed behavior happens every time. Just the timing is different and the time from starting the program to hearing the monsterish voice is between 3 and 10 seconds. It is also rather strange that this time span can be so huge!
>
>
> What happens here?! In its current form the SIPPhone is pretty unusable. I hope you have ideas how to fix that :) I have experimented with many of the MediaConfig parameters with no success.
>
>
> Thank you!
>
> Cheers,
> Oliver
>
>
>
> _______________________________________________
>
> Visit our blog: http://blog.pjsip.org
>
>
>
> pjsip mailing list
>
> pjsip at lists.pjsip.org<mailto:pjsip at lists.pjsip.org>
>
> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
>
> _______________________________________________ Visit our blog: http://blog.pjsip.org pjsip mailing list pjsip at lists.pjsip.org http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
>
>
> _______________________________________________
>
> Visit our blog: http://blog.pjsip.org
>
>
>
> pjsip mailing list
>
> pjsip at lists.pjsip.org
>
> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
>
> _______________________________________________ Visit our blog: http://blog.pjsip.org pjsip mailing list pjsip at lists.pjsip.org<mailto:pjsip at lists.pjsip.org> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
> __________________________________________________________________________________
> SELECTRIC Nachrichten-Systeme GmbH
> Haferlandweg 18
> 48155 Muenster
>
> Internet:
> www.selectric.de
>
> ---
> Sitz der Gesellschaft (Company Premises): Muenster | Registergericht (Register Court): Amtsgericht Muenster | HRB-Nr. 1264
> Geschaeftsfuehrer (General Managers): Michael Heussner, Hendrik Pieper
>
>
> _______________________________________________
> Visit our blog: http://blog.pjsip.org
>
> pjsip mailing list
> pjsip at lists.pjsip.org
> http://lists.pjsip.org/mailman/listinfo/pjsip_lists.pjsip.org
>
> -- 
> 1024D/A9F85A52      2000-01-18      Alain Totouom <totouom at gmx.de>
> PGP Fingerprint DA18 0DF2 FBD2 5F67 0656 452D E3A2 7531 A9F8 5A52
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.pjsip.org/pipermail/pjsip_lists.pjsip.org/attachments/20160407/e8f7b9c5/attachment-0002.html>


More information about the pjsip mailing list