Speex dev - Dec 2007 - Echo cancellor on Windows, a few questions.

On Dec 3, 2007 1:24 PM, Stefan Reuther <streu@gmx.de>
wrote:
> John Miles wrote:
> > What's wrong with a plain old .bat file, or even an NMAKE .mak
file?
> > Ship two files, debug.bat and release.bat, and call it good.
> >
> > It is best to leave project-file creation up to individual users,
> > in my opinion.
>
> I second that. When I played around with Speex a while ago, I just
> dumped the Speex source files into a .dsp file (Visual Studio Project),
> and built that. For another environment, I made a 25 line Makefile
> (essentially, a SOURCE = ..... line plus the auto-dependency example
> code from the GNU make manual).
>
> The only "problem" is to find out what files are needed for the
library
> and which are superfluous (i.e. test/demo programs with their own
> main()). A minimal Makefile would help to document that, and enable
> people create projects, with minimum effort, for all sorts of
> environments nobody ever heard of.
For those people building on Windows, please use the automake/autoconf
build system as your reference. The .c files needed to build the speex
libraries are spelled out quite concisely and explicitly in
libspeex/Makefile.am.

Creating yet another build system for speex would, IMHO, only serve to
obfuscate and cause yet more build-related chatter on this list.

Pete

Peter Grayson wrote:
> On Dec 3, 2007 1:24 PM, Stefan Reuther <streu@gmx.de> wrote:
>>The only "problem" is to find out what files are needed for
the library
>>and which are superfluous (i.e. test/demo programs with their own
>>main()). A minimal Makefile would help to document that, and enable
>>people create projects, with minimum effort, for all sorts of
>>environments nobody ever heard of.
> 
> For those people building on Windows, please use the automake/autoconf
> build system as your reference. The .c files needed to build the speex
> libraries are spelled out quite concisely and explicitly in
> libspeex/Makefile.am.
Agreed. That's exactly what one needs to build such a project. I wonder
why I hadn't found that at that time. Case closed.


  Stefan

what i want to achieve is simple, i want to be able to sitback in my chair 
and talk over a voip call using speex. as the mic and the speakers are 
built into the laptop, an echo canceller is required.

what i am doing is this :

First, each time i finish writing a block to the waveout, i copy it into a 
buffer like this:
 	Segment *pseg = (Segment *)phead;
 	waveOutUnprepareHeader(waveOut, phead, sizeof(WAVEHDR));
 	memcpy(remote, pseg->samples, 160 * sizeof(short));

Second, each time i get samples from the mic, i do the echo cancellation 
like this:
 	if (nsamples == 160){
 		speex_echo_cancellation(echo_state, (short *)pseg->samples, remote,
unechoed);
 		q = unechoed;
 	}

I have also initialized the echo cancellor like this:
 	echo_state = speex_echo_state_init(160, 8000);

My issue is simply that the echo cancelleor is making almost no difference 
to my echo at all. any ideas?

Q1. is there a way to determine if the echo cancellor managed to lock in or not?

Q2. does anybody have a working code with wave apis?

Q3. is the speech processor necessary for the echo cancellor to work? what 
is the difference between 'echo' and the 'residual echo' that
the
preprocessor is suppoed to cancel?

  - farhan