iaxclient

At astricon, some iaxclient users had talked with me about a problem 
they have where clients would have poor audio quality (from microphone), 
when using iaxclient, because their mixer settings were too high, and 
the incoming audio was clipping.

This is a patch which needs further testing, which implements a loose 
mixer-based AGC method to iaxclient.  Basically the way it works is that 
when iaxclient detects speech, it looks at the signal level, and makes 
adjustments to the mixer levels.  The adjustment is slow, and has a high 
level of hysteresis, so that once it finds the proper position, it 
shouldn't move, and fine adjustments will be done with multiplication.  
[otherwise, the other DSP functions, like denoising, would get confused 
by the changing input levels].


Index: audio_encode.c
===================================================================
RCS file: /cvsroot/iaxclient/iaxclient/lib/audio_encode.c,v
retrieving revision 1.18
diff -u -r1.18 audio_encode.c
--- audio_encode.c      25 Jun 2004 19:38:59 -0000      1.18
+++ audio_encode.c      2 Aug 2004 23:09:23 -0000
@@ -86,6 +86,29 @@
     /* only preprocess if we're interested in VAD, AGC, or DENOISE */
     if((iaxc_filters & (IAXC_FILTER_DENOISE | IAXC_FILTER_AGC)) ||
iaxc_silence_threshold > 0)
        silent = !speex_preprocess(st, audio, NULL);
+    /* bring speex AGC gain out to mixer */
+    if(!silent && (iaxc_filters & IAXC_FILTER_AGC )) {
+       static int i;
+       double level;
+       i++;
+
+       if((i&0x3f) == 0) {
+           /* fprintf(stderr, "loudness = %f\n", st->loudness2); */
+           level =  iaxc_input_level_get();
+           /* lower quickly if we're too hot */
+           if(st->loudness2 > 20000 && level > 0.3) {
+             level -= 0.2;
+             iaxc_input_level_set(level);
+           } /* lower slowly */
+           else if(st->loudness2 > 8000) {
+              if(level >= 0.1) iaxc_input_level_set(level - 0.1);
+           }
+           /* raise slowly if we're cold */
+           else if( st->loudness2 < 4000) {
+                if(level <= 0.9) iaxc_input_level_set(level + 0.1);
+           }
+       }
+    }


Comments on this are welcome.

The one thing that should probably be added with this is to set the 
input volume up to the max on startup. Because if the input levels are 
too low to start with, speech won't ever be detected, and the AGC won't 
have any effect.

-SteveK

Hi,

Does anyone know if iaxclient and related libraries could work on Windows
95?

Thanks,
Alex

> -----Original Message-----
> From: iaxclient-devel-admin@... 
> [mailto:iaxclient-devel-admin@...] On 
> Behalf Of Steve Kann
> Sent: Wednesday, September 29, 2004 9:19 PM
> To: Ilguiz Latypov
> Cc: iaxclient-devel@...
> Subject: Re: [Iaxclient-devel] active development?
> 
> 
> On Sep 29, 2004, at 9:38 PM, Ilguiz Latypov wrote:
> 
> >
> > On Wed, Sep 29, 2004 at 09:57:59PM -0300, antonio@...
> > wrote:
> >> Or everyone that is working with this project keep the new code 
> >> private and forget about contribute back? What happens?

> snip <

> The people who have contributed so far, are basically 
> occupied with other things.  The code here works, so we've 
> all scratched our immediate itches, so to speak.

Speaking for myself, that's exactly it. I haven't made any changes to
iaxclent in months.

I am starting to feel the codec itch, however.

---
Outgoing mail is certified Virus Free.
Checked by AVG anti-virus system (http://www.grisoft.com).
Version: 6.0.771 / Virus Database: 518 - Release Date: 9/28/2004

On Sep 29, 2004, at 5:04 PM, Geoff Nordli wrote:

>> --- NEW FILE: README ---
>> app_conference is a channel-independent conference application.
>>
>> Design goals
>> It has several design goals which are different than meetme:
>>
>>     * It does not require a zap channel for timing.
>>     * It is very efficient when used with channels which
>> support DTX (silence
>> 	detection/discontinuous transmission).
>>
>>     * It can do VAD on channels which do not support DTX (although
>> 	this is probably more expensive than just mixing them).
>>     * It presents messages on the Monitor interface for
>> determine which
>> 	speakers are active.
>>
>
>
> Hi Steve.
>
> Does it have the ability to mute/unmute users during a conference?
> Something like the app_meetme2 functionality.

Not presently, but we might add that sometime.

You can mute/unmute from the manager interface, by transferring the 
call back into the conference with different options.

-SteveK

Is there anyone working in this project??? Actively???

As far I can see... there is few messages in the list...
Very very few about improving the library (about bugs, new features, new
codecs...)
Few answers...
No changes in the cvs....

Trying to understand the situation... Maybe this lib is not so modular or
featured as asterisk itself, but it is a good piece of code... It works
very well... Is there some other (better?) iax library that I don't know
about? Or everyone that is working with this project keep the new code
private and forget about contribute back? What happens?

[]'s
Moreiras.




--------------------------------------------------
               Email Plus Estadão 
Agora você pode ter mais ferramentas e espaço para 
armazenar seus emails. 
http://www.estadao.com.br/webmail/pago/ 
--------------------------------------------------

Hello.

There is a lot of posts about compiling iaxclient under vc++ 6.0 but I can't
find information how to solve my problem. Somehow I managed to compile
iaxclient_lib but when I try to compile iaxclient using that library i get
linking problems like:

-------------------Configuration: WinIAX - Win32 Debug--------------------
Linking...
iaxclient_lib.lib(iaxclient_lib.obj) : error LNK2001: unresolved external
symbol _iax_transfer
iaxclient_lib.lib(iaxclient_lib.obj) : error LNK2001: unresolved external
symbol _iax_destroy
iaxclient_lib.lib(iaxclient_lib.obj) : error LNK2001: unresolved external
symbol _iax_quelch_moh
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _speex_echo_cancel
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _speex_echo_state_init
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _speex_echo_state_destroy
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol __Px_SetMicrophoneBoost
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol __Px_SetCurrentInputSourceByName
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_OpenMixer
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_CloseMixer
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_GetInputVolume
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_GetNumInputSources
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_GetOutputVolume
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_SupportsPCMOutputVolume
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_SetInputVolume
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol _Px_SetOutputVolume
iaxclient_lib.lib(audio_portaudio.obj) : error LNK2001: unresolved external
symbol __Px_GetMicrophoneBoost
Debug/WinIAX.exe : fatal error LNK1120: 17 unresolved externals
Error executing link.exe.

WinIAX.exe - 18 error(s), 0 warning(s)

I use source downloaded from http://iaxclient.sourceforge.net/iaxcomm/

I did a Complete Re Write if IAX Client Based on Asterisk Changes and its
Running Perfect and I am my using CygWin or MinGW I am Using Visual Studio
.Net 2003 Enterprise.

When I finish the last of the changes this week I will post on News Group
letting people know where they can download it.

I'm here at Astricon 2004 already.

Anyone else coming or here yet?

e-mail me, and we can get together.

-SteveK

Hi,

First, let me introduce myself... I'm Moreiras, from Brazil, new to this
list, and new to the "asterisk" word. I'm working with asterisk for two
months and with iaxclient and iaxcomm (in Windows) for a little more than
1 month.  I think I'm a newbe in this hole thing.

In time, I think my english is not very good. But I will try to be clear...

I have three questions about the library, that I'm going to ask in
separated emails.

First one: I have sound problems in this configuration:

IAXCOMM Windows (GSM) ---> Asterisk Server ---> Cisco SIP IP Phone (ULAW)

The sound quality in IAXCOMM side is OK. But at the Cisco side, it is
horrible. Looking in asterisk email lists, bugs, and google I found that
probably it is a problem with generated IAX2 timestamps. The problems is
already resolved in asterisk server. But maybe it is not in Iaxcomm /
iaxclient library, I don't know for sure.

The timestamps generated by iaxcomm are a little erratic. There are
increments between 15ms and 30ms between the timestamps of consecutive
packets. (One can see that with tcpdump, for example). The asterisk server
gets that timestamps then generate the SIP timestamps... And the CISCO
phone doesn't like erratic timestamps, it drop some packets, and the sound
quality go down...

The following configuration sounds OK:

IAXCOMM (GSM)  ---->  Asterisk Server Meeetme app <---- Cisco SIP (ULAW)

I think that in this configuration the server generate the timestamps
without look at that generated by iaxcomm...  everything works fine.

At iax.c, in the library (libiax2), I see that there is some code to
generate the timestamps based on the timing from the audio sampling
(8khz). The code is not used by default, the variable
USE_VOICE_TS_PREDICTION have to be defined to do this.

Well, I put a:
#define USE_VOICE_TS_PREDICTION 1
at the beginning  of iax.c.

Everything works fine, then. Time difference between timestamps is
something more than 20ms, near 23ms, but it is ever 23ms, no more erratic
differences, and it seams to don't generate sound problems. And the sound
at the SIP ULAW CISCO side is OK, now.

Well...

I'd like to know if there is a problem that I don't see with this code
(VOICE_TS PREDICTION). Maybe at other plataforms (I'm using windows...)?
If no, why this is not the default choice?

At asterisk lists I see that this problem (erratic timestamps with IAX2
and GSM) was resolved in asterisk, but I was not able to identify this
correction looking at the code (not so good programmer... hehehe) Is there
some other way to do this?

Thanks!

Moreiras.







--------------------------------------------------
Estadão - Internet com alta qualidade de conexão.
GANHE ACESSO GRATUITO à Internet do Estadão em 
http://www.estadao.com.br/discador/
--------------------------------------------------

The poke_reply is not implemented at libiax2, even at cvs head... Is there
a special reason for this?

I've implemented this here, and now qualifying iaxcomm users is possible,
I think it helps me to test the connection quality in some situations, and
I see no reason to not do this...

asterisk*CLI> iax2 show peers
Name/Username    Host                 Mask             Port      Status
4596/Jara        172.17.22.200   (D)  255.255.255.255  4569      OK (15 ms)
2520/Hideo       (Unspecified)   (D)  255.255.255.255  0         UNKNOWN
3569/Simao       172.21.35.83    (D)  255.255.255.255  4569      OK (4 ms)
(...)
(4596 and 3569 are iaxcomm clients)

You think there is a problem implementing this at libiax2? and qualifying
Iax clients?

What I did here:

*** at iax.c (libiax2)

** added near line 962
static int iax_send_poke_reply(struct iax_session *session, unsigned int ts);

** added near line 989:
			case IAX_EVENT_POKE:
				event->etype = IAX_EVENT_PONG;
				iax_send_poke_reply(event->session, event->ts);
				iax_event_free(event);
				break;

** added near line 1142
static int iax_send_poke_reply(struct iax_session *session, unsigned int ts)
{
	return send_command_final(session, AST_FRAME_IAX, IAX_COMMAND_PONG, ts,
NULL, 0, -1);
}


I think it is all...  Let me know if I forget something... Or if it was
made the wrong way... It's working here, then, if it not work for you,
maybe I forget to put something in this mail... I can try generate a patch
if someone became interested, but I think it's not necessary, the idea is
very simple....

Thanks and []'s

Moreiras.



--------------------------------------------------
Estadão - Internet com alta qualidade de conexão.
GANHE ACESSO GRATUITO à Internet do Estadão em 
http://www.estadao.com.br/discador/
--------------------------------------------------

Sometimes Iaxcomm miss the incoming call...

I found a patch at
http://www.nexusuk.org/projects/VoIP/iaxcomm/missedcalls
that resolves the problem (or seams to).

Basicaly... at iax.c (libiax2)

static int send_command_final(struct iax_session *i, char type, int
command, unsigned int ts, char *data, int datalen, int seqno)
{

//	return __send_command(i, type, command, ts, data, datalen, seqno, 0, 0,
1, 0);

	int r;
	r = __send_command(i, type, command, ts, data, datalen, seqno, 0, 0, 1, 0);
	if (r >= 0) destroy_session(i);
	return r;

}

I'm not working with the cvs head... but with the version at
http://iaxclient.sourceforge.net/snapshots/iaxclient.tar.gz.

But it seams that there is nothing new adressing this problem at cvs
head... or there is and I didn't see?

Or is it a problem at the iaxcomm addressed from the wrong way with that
patch?

Thanks.

Moreiras



--------------------------------------------------
Estadão - Internet com alta qualidade de conexão.
GANHE ACESSO GRATUITO à Internet do Estadão em 
http://www.estadao.com.br/discador/
--------------------------------------------------

Using the jabra-250 bluetooth headset for the iaxclient doesn't work, 
as it is only a mono device, and "audio_portaudio.c" requires stereo.

However, if I simply comment out the "#ifndef MACOSX" on line 449, 
it works, as does the internal mic and external speakers. Am I missing
something, or can the "#ifndef MACOSX" be safely eliminated?

S

Eric Alexander wrote:

>Ah.. Please excuse my ignorance. Can you point me in the right direction to
>install app_conference? I have tried installing it on 3 boxes (debian,
>slackware, redhat) and make install fails with the following everytime. 
>
>cc1: Invalid option `abi=altivec'
>cc1: Invalid option `dynamic-no-pic'
>cc1: unknown C standard `c99'
>cc1: Invalid option `-fprefetch-loop-arrays'
>cc1: Invalid option `-fsingle-precision-constant'
>cc1: Invalid option `-faltivec'
>cc1: bad value (7450) for -mcpu= switch
>app_conference.c: In function `add_milliseconds':
>app_conference.c:137: parse error before `long'
>app_conference.c:140: `s' undeclared (first use in this function)
>app_conference.c:140: (Each undeclared identifier is reported only once
>app_conference.c:140: for each function it appears in.)
>  
>
It appears that the current Makefile is set up to compile under Darwin/PPC.

Just comment out the third CFLAGS line, and if you're using an old GCC, 
remove the -std=c99 option from the first CFLAGS line.

-SteveK



> 
>
>-----Original Message-----
>From: Steve Kann [mailto:stevek@...] 
>Sent: Thursday, September 09, 2004 9:36 AM
>To: Eric Alexander
>Cc: iaxclient-devel@...
>Subject: Re: [Iaxclient-devel] app_conference
>
>Eric Alexander wrote:
>
>  
>
>>Is the CVS missing app_conference.o ?
>> 
>>
>>    
>>
>No, object files don't normally go into CVS repositories..
>
>-SteveK
>
>  
>