[ previous ] [ next ] [ threads ]
 To :  "G.Jacobsen" <g_jacobsen@y...>
 From :  Nik Pakar <nikpakar@g...>
 Subject :  Re: [yate] Re: Codec packet size in both ways
 Date :  Sat, 17 Oct 2009 11:57:22 +0530
Hi Gerry,

I have done a mistake earlier by not having g729 on modules. Thats what that
error gave me.

Now i have compiled and got g729.yate on it.

Now if i set that g729, 40, 40000 on DataFormat.cpp, i get following WARN
when starting yate and g729 get disabled. Only if i set it to 10, 10000 it
works.

This is what i get when set to 40, 40000

Loaded module G.729a - based on Intel IPP
 Tried to register 'audio' format 'g729' fsize=10 ftime=10000
srate=8000 nchan=1

Any idea on where this original fsize and ftime comes from ?

Thanks for the help.
Rgds
Nik

On Wed, Oct 14, 2009 at 11:23 PM, Nik Pakar  wrote:

> Hi Gerry,
> Yes, its purely to save bandwidth. 20ms uses almost double. :(
>
> Nik.
>
>
> On Wed, Oct 14, 2009 at 11:21 PM, G.Jacobsen wrote:
>
>>  Nik,
>>
>> I didnt check whether it actually works on 20ms :-)
>>
>> All we did was look into the yate header file where these parameters are
>> defined and the G729 docs and hacked it.
>>
>> And it works with all gateways I have encountered so far.
>>
>> Why do you want to use 40ms, save bandwidth ?
>>
>> Cheers
>>
>> Gerry
>>
>>
>>
>>
>>
>>
>>
>>
>> ----- Original Message -----
>>
>> *From:* Nik Pakar 
>> *To:* G.Jacobsen 
>> *Cc:* yate@v...
>> *Sent:* Wednesday, October 14, 2009 8:24 PM
>> *Subject:* Re: [yate] Re: Codec packet size in both ways
>>
>> Hi Gerry, by setting it 20,20000, is it working at 40ms sampling now ?
>> Or are you doing it on 20ms ?
>>
>> Rgds
>> Nik
>>
>> On Wed, Oct 14, 2009 at 10:47 PM, G.Jacobsen wrote:
>>
>>>  Hi Nik,
>>>
>>> "Also if that 10, 10000 was effective, how could it was acting as 20ms by
>>> default "
>>>
>>> Indeed an interesting question.
>>>
>>> BTW: We have that on 20,20000 on our live server and that has worked for
>>> months without problems. Reading through the G729 documentation my
>>> understanding was that 20,20000 is the standard so we hacked that setting.
>>>
>>> Gerry
>>>
>>>
>>>
>>> ----- Original Message -----
>>>
>>> *From:* Nik Pakar 
>>> *To:* Gerrit Jacobsen 
>>> *Cc:* yate@v...
>>> *Sent:* Wednesday, October 14, 2009 6:22 PM
>>> *Subject:* Re: [yate] Re: Codec packet size in both ways
>>>
>>> Hey gerry, many thanks for the hint.
>>> I just tried it. But calls with no audio at all. and the logs shows
>>> below.
>>>
>>>  [0m [K [40;1;36m [K No DataTranslator created for 'g729' ->
>>> 'alaw'
>>> [0m [K [40;1;31m [K DataTranslator::attachChain [0xb7812950] 'g729'
>>> -> [0x8344440] 'alaw' failed
>>>
>>> Would there be any other places also to change.
>>>
>>> Also if that 10, 10000 was effective, how could it was acting as 20ms by
>>> default :(
>>>
>>> Once again many thanks and appreciate your help.
>>>
>>> Rgds
>>> Nik
>>>
>>> On Wed, Oct 14, 2009 at 11:18 AM, Gerrit Jacobsen <
>>> g_jacobsen@y...> wrote:
>>>
>>>> Nik,
>>>>
>>>> I would guess this can be hacked in
>>>>
>>>> DataFormat.cpp
>>>>
>>>> Try
>>>> FormatInfo("g729", 40, 40000),
>>>>
>>>> Gerry
>>>>
>>>>
>>>>
>>>>
>>>> --- On Tue, 13/10/09, Nik Pakar  wrote:
>>>>
>>>> > From: Nik Pakar 
>>>> > Subject: [yate] Re: Codec packet size in both ways
>>>> > To: "Yate mailing list" 
>>>> > Date: Tuesday, 13 October, 2009, 9:51 AM
>>>>  > Hello friends,
>>>> > Can any one please give me some hint on
>>>> > this.
>>>> > When using g729 type codec, call orginating yate
>>>> > server send the ptime as 40. But the receiving end
>>>> > yate's codec encoder talk back in 20ms. How can we set
>>>> > this to be on 40ms from both side.
>>>> >
>>>> > Thanks a lot for any help.
>>>> > RgdsNik
>>>> >
>>>> > On Wed, Oct 7, 2009 at 6:09 AM,
>>>> > Nik Pakar 
>>>> > wrote:
>>>> >
>>>> > Hello friends,
>>>> >
>>>> > I have yate server 1 sending a call to yate server 2.
>>>> > I have set osdp_ptime=40 on the yate1. The call originate
>>>> > with ptime=40. But when the call is live, and media stream
>>>> > established. i can see the call path
>>>> >
>>>> > yate 1 -> yate 2 = is on 40ms
>>>> > yate 2 -> yate 1 = is on 20ms
>>>> >
>>>> > How and where can i set this reverse path packet size
>>>> > for codecs. Since there is no regex involved on the
>>>> > receiving yate2 server for this call, is there any global
>>>> > parameter to set for this.
>>>> >
>>>> > Appreciate any help.
>>>> >
>>>> > Thanks
>>>> > nik
>>>> >
>>>> >
>>>> >
>>>>
>>>>
>>>>
>>>>
>>>>
>>>
>>
>



Hi Gerry,

I have done a mistake earlier by not having g729 on modules. Thats what that error gave me.

Now i have compiled and got g729.yate on it.

Now if i set that g729, 40, 40000 on DataFormat.cpp, i get following WARN when starting yate and g729 get disabled. Only if i set it to 10, 10000 it works.

This is what i get when set to 40, 40000

Loaded module G.729a - based on Intel IPP
<WARN> Tried to register 'audio' format 'g729' fsize=10 ftime=10000 srate=8000 nchan=1

Any idea on where this original fsize and ftime comes from ?

Thanks for the help.
Rgds
Nik

On Wed, Oct 14, 2009 at 11:23 PM, Nik Pakar <nikpakar@g...> wrote:
Hi Gerry,

Yes, its purely to save bandwidth. 20ms uses almost double. :(

Nik.


On Wed, Oct 14, 2009 at 11:21 PM, G.Jacobsen <g_jacobsen@y...> wrote:
Nik,
 
I didnt check whether it actually works on 20ms :-)
 
All we did was look into the yate header file where these parameters are defined and the G729 docs and hacked it.
 
And it works with all gateways I have encountered so far.
 
Why do you want to use 40ms, save bandwidth ?
 
Cheers
 
Gerry
 
 
 
 
 
 
 
 
----- Original Message -----
Sent: Wednesday, October 14, 2009 8:24 PM
Subject: Re: [yate] Re: Codec packet size in both ways

Hi Gerry, by setting it 20,20000, is it working at 40ms sampling now ?

Or are you doing it on 20ms ?

Rgds
Nik

On Wed, Oct 14, 2009 at 10:47 PM, G.Jacobsen <g_jacobsen@y...> wrote:
Hi Nik,
 
"Also if that 10, 10000 was effective, how could it was acting as 20ms by default "
 
Indeed an interesting question.
 
BTW: We have that on 20,20000 on our live server and that has worked for months without problems. Reading through the G729 documentation my understanding was that 20,20000 is the standard so we hacked that setting.
 
Gerry
 
 
 
----- Original Message -----
From: Nik Pakar
Sent: Wednesday, October 14, 2009 6:22 PM
Subject: Re: [yate] Re: Codec packet size in both ways

Hey gerry, many thanks for the hint.

I just tried it. But calls with no audio at all. and the logs shows below.

[0m [K [40;1;36m [K<INFO> No DataTranslator created for 'g729' -> 'alaw'
[0m [K [40;1;31m [K<WARN> DataTranslator::attachChain [0xb7812950] 'g729' -> [0x8344440] 'alaw' failed

Would there be any other places also to change.

Also if that 10, 10000 was effective, how could it was acting as 20ms by default :(

Once again many thanks and appreciate your help.

Rgds
Nik

On Wed, Oct 14, 2009 at 11:18 AM, Gerrit Jacobsen <g_jacobsen@y...> wrote:
Nik,

I would guess this can be hacked in

DataFormat.cpp

Try
FormatInfo("g729", 40, 40000),

Gerry




--- On Tue, 13/10/09, Nik Pakar <nikpakar@g...> wrote:

> From: Nik Pakar <nikpakar@g...>
> Subject: [yate] Re: Codec packet size in both ways
> To: "Yate mailing list" <yate@v...>
> Date: Tuesday, 13 October, 2009, 9:51 AM
> Hello friends,
> Can any one please give me some hint on
> this.
> When using g729 type codec, call orginating yate
> server send the ptime as 40. But the receiving end
> yate's codec encoder talk back in 20ms. How can we set
> this to be on 40ms from both side.
>
> Thanks a lot for any help.
> RgdsNik
>
> On Wed, Oct 7, 2009 at 6:09 AM,
> Nik Pakar <nikpakar@g...>
> wrote:
>
> Hello friends,
>  
> I have yate server 1 sending a call to yate server 2.
> I have set osdp_ptime=40 on the yate1. The call originate
> with ptime=40. But when the call is live, and media stream
> established. i can see the call path
>  
> yate 1 -> yate 2 = is on 40ms
> yate 2 -> yate 1 = is on 20ms
>  
> How and where can i set this reverse path packet size
> for codecs. Since there is no regex involved on the
> receiving yate2 server for this call, is there any global
> parameter to set for this.
>  
> Appreciate any help.
>  
> Thanks
> nik
>
>
>