[Speex-dev] AEC Patented ?

Jean-Marc Valin jean-marc.valin at usherbrooke.ca
Tue Oct 3 14:30:15 PDT 2006

Don't worry. Not only does the method seems different, but it also only
applies to NLMS, while me EC is using MDF, which is a frequency-domain
echo canceller.


Jean-Christophe.Berge at etu.enseeiht.fr a écrit :
> Hello Jean-Marc Valin and all the Speex-dev team,
> Today I have a little question about patents and the AEC.
> I' am not a lawyer, but after a deep research in esp at cenet.com,
> I found that in August 2003, an AEC called "Double talk insensitive NLMS
> algorithm" (DNLMS) have been patented by Gary Q Jin, Gordon J Reesor, and
> others (US 6.608.897 B1).
> It uses a continuous adaptation step as speex, to avoid adaptatif filter
> divergence during double-talk.
> Is the computation method really the same as speex AEC (hope not) ?
> Somebody who have a deep understanding of Speex AEC algorithm might have a
> look at this patent ?
> Thank you in advance for your answer,
> Sincerly
> Jean-Christophe
> _______________________________________________
> Speex-dev mailing list
> Speex-dev at xiph.org
> http://lists.xiph.org/mailman/listinfo/speex-dev

More information about the Speex-dev mailing list