Re: [GnomeMeeting-devel-list] Packet loss masking

Am Mittwoch, 6. April 2005 16:25 schrieb Damien Sandras:
> Le mercredi 06 avril 2005 à 16:21 +0200, Dalibor Budimir a écrit :
> > Hello,
> >
> > (I need this for school project: masking handoff loss in Mobile IPv6
> > network)
> >
> > When there is loss in audio stream, I should repeat last receved half
> > second of audio.
Just repeating the last half second is not a good idea in general. You should 
repeat the last sound only (search for linguistic and (fricative, plosive, 
formant)), else you will get a really anoying stuttering noise.

> > I am planing to use Gnomemeeting over openh323.
> >
> > Is it enough to do some changes in sound_handling.cpp in Gnomemeeting?
> unfortunately not, you will have to change things in openh323, but I
> don't know exactly where.
> I think the SPEEX codec has the feature you are describing, so it could
> be a the codec level.

The codec level is the easy place to do this - speech codecs are based on a 
linguistic model, so it is much easier to get the last sound. On the other 
hand it can be done independent of the codec. 

> > Any sugestions a/o solution are welcome -- I am beginner.

Search for papers about packet loss concealment for voice. Google gives me 
30000 hits, and the most look promising, although some may lack technical 



Stefan Brüns  /  Kastanienweg 6 - Zimmer 1206  /  52074 Aachen
mailto:lurch gmx li
   phone: +49 241 169-4206     mobile: +49 160 3797725

Attachment: pgp1aET5dWp8A.pgp
Description: PGP signature

[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]