[Speex-dev] Problems with Speex echo canceller in 1.1.10

[Eric Scott]

Jean-Marc,

BTW, the problem #1 is not unique to the speex echo canceller.  It has also occurred on other ec's I've tried, most recently the CANEC ec from dspalgorithms.com.

- Eric Scott

> -----Original Message-----
> From: Eric Scott 
> Sent: Thursday, July 28, 2005 9:11 AM
> To: 'Jean-Marc Valin'
> Cc: speex-dev at xiph.org
> Subject: RE: [Speex-dev] Problems with Speex echo canceller in 1.1.10
> 
> 
> Jean-Marc,
> 
> That was my first thought as well.  However, I double checked 
> everything and it wasn't reversed.  I also reversed it to see 
> what would happen.  In that case, the NES came out unchanged. 
>  (This NES was the original FES.)
> 
> Any other ideas?  Also, do you know what algorithm is used?
> 
> - Eric Scott
> 
> > -----Original Message-----
> > From: Jean-Marc Valin [mailto:Jean-Marc.Valin at USherbrooke.ca]
> > Sent: Wednesday, July 27, 2005 8:43 PM
> > To: Eric Scott
> > Cc: speex-dev at xiph.org
> > Subject: Re: [Speex-dev] Problems with Speex echo canceller 
> in 1.1.10
> > 
> > 
> > Le mercredi 27 juillet 2005 à 12:57 -0700, Eric Scott a écrit :
> > > I'm having trouble getting the echo canceller to work and 
> > I'm hoping some one can get me started in diagnosing what's 
> > going on.  The symptoms I'm seeing are these:
> > > 
> > > 1. If the NES is silence (i.e., no echo, no speech), the ec 
> > puts the FES back into the audio.
> > > 2. If the NES is similar to the FES (same speech words), 
> > the ec doesn't seem to remove the echo.
> > 
> > With just this, I can tell that you simply inverted the near 
> > and far end
> > signals.
> > 
> > 	Jean-Marc
> > 
> > -- 
> > Jean-Marc Valin <Jean-Marc.Valin at USherbrooke.ca>
> > Université de Sherbrooke
> > 
> > 
>