[asterisk-dev] Jitter Buffer

Pavel Jezek pavel.jezek at i.cz
Wed Oct 25 01:51:17 MST 2006


zoachien at securax.org wrote:
>
> And that is where forcejb comes in :)
> The phones usually have a max jitter buffer of 128ms, if you do 
> forcejb, you could make this bigger.

yes, I would like to do this, but it doesn't work!
I never see message creating jitterbuffer on BOTH incomming call legs 
e.g. SIP/xxx and skinny/xxx
sound is very jerky with many clicks when jitter occur
I have jbenable & jbforce on sip.conf & skinny.conf, so it is bug or no?!
PJ



>
> Zoa
>
> Pavel Jezek wrote:
>>
>> zoachien at securax.org wrote:
>>> We should not dejitter when possible, we should dejitter when we 
>>> need to because the receiving endpoint cannot do dejittering.
>>> Sip to iax should not dejitter, as the endpoints should dejitter,
>>
>> I'm testing with ci$co 7912/sip and ci$co 7920/skinny, both phones 
>> using codec g729 and are connected over cdma connection (with jitter 
>> ~1000ms - mainly in upload direction i.e. phone->asterisk), asterisk 
>> is in the midle, sound quality is still very poor  :'(
>> maybe I have bad phones for this use (maybe with too small internal 
>> jitterbuffer), but I think, that proper dejittering at asterisk for 
>> both legs incomming from phones should help...
>> PJ
>>
>>
>>
>>
>>> the reason for this is that if we do dejittering in the middle, and 
>>> another time on the endpoints, the total delay because of 
>>> dejittering will be bigger than needed.
>>>
>>> A configuration file for applications might make sense, for example 
>>> if you do recordings because of line quality checks, you don't want 
>>> the jitter buffer to be active, as you want the real thing with the 
>>> gaps etc, but if you want recordings for monitoring people, you will 
>>> want it because it will be nicer to listen to.
>>>
>>> Zoa
>>>
>>>
>>> Pavel Jezek wrote:
>>>> Zoa, I think, that we should dejjitter when possible, always 
>>>> _before_ any processing within asterisk (sending to some app, 
>>>> bridging to channel, etc.),
>>>> is there any reason, why (in your example) dejjitter only sip-zap 
>>>> and not sip-iax?
>>>> PJ
>>>>
>>>>
>>>>
>>>> zoachien at securax.org wrote:
>>>>>
>>>>>> Unless we're planning on adding jbenable-type options to 
>>>>>> voicemail, meetme, monitor/mixmonitor and local channels (as a 
>>>>>> starting point), it would seem to make a LOT more sense to keep 
>>>>>> jbenable in all voip protocol configurations... (i.e. sip, iax2, 
>>>>>> skinny, etc.) -- that's where the jitter buffer actually exists.
>>>>>>
>>>>>> -A.
>>>>>>   
>>>>>
>>>>> I disagree with this, it's the only way you can choose when to 
>>>>> have dejittering and when not.
>>>>> You might for example want it for sip to zap but not from sip to 
>>>>> iax, thus if you set it in sip, you will have dejittering for sip 
>>>>> in your suggestion.
>>>>>
>>>>> Zoa
>>>>> _______________________________________________
>>>>> --Bandwidth and Colocation provided by Easynews.com --
>>>>>
>>>>> asterisk-dev mailing list
>>>>> To UNSUBSCRIBE or update options visit:
>>>>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>>>> _______________________________________________
>>>> --Bandwidth and Colocation provided by Easynews.com --
>>>>
>>>> asterisk-dev mailing list
>>>> To UNSUBSCRIBE or update options visit:
>>>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>>>
>>> _______________________________________________
>>> --Bandwidth and Colocation provided by Easynews.com --
>>>
>>> asterisk-dev mailing list
>>> To UNSUBSCRIBE or update options visit:
>>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>> _______________________________________________
>> --Bandwidth and Colocation provided by Easynews.com --
>>
>> asterisk-dev mailing list
>> To UNSUBSCRIBE or update options visit:
>>   http://lists.digium.com/mailman/listinfo/asterisk-dev
>
> _______________________________________________
> --Bandwidth and Colocation provided by Easynews.com --
>
> asterisk-dev mailing list
> To UNSUBSCRIBE or update options visit:
>   http://lists.digium.com/mailman/listinfo/asterisk-dev


More information about the asterisk-dev mailing list