First, note that what I say here applies to any speech codec (not just
Speex). There is no simple solution to that problem. In fact, there are
two sources of quality degradation. The first is the tandeming, which
means encoding and decoding twice. The second, and more important
problem is that the second time you will encode the sum of several
signals. That's a problem for most (all) speech codecs, as they are
designed for only a single voice. When more there one voice is present,
the quality is likely to drop quickly. The only way to counter that
would be to use really high bit-rates (like 18.2 or 24.6) for the sum.
The other solution is to transmit the streams separately. Since there
are usually at most two persons talking at the same time, it probably
wouldn't add that much bandwidth.
Of course, it all depends on the bit-rate you want to use. For low
bit-rates (e.g. 8 kbps), it's probably better to transmit the streams
separately, while for higher bit-rates (e.g. 15 kbps), re-encoding at
24.6 kbps might be the best solution.
Jean-Marc
> I was wondering what the theoretical ramifications where of re-encoding an
> already decoded Speex stream. This would be used to mix a live stream of
> several Speex end-points for a group-conferencing product?
>
> Would there be a significant impact on the quality - Should I set bit-rate
> higher on the re-encode side? Or is there a fancy way to mix the encoded
> samples?
>
> BTW - Thanks for a great codec, We are using Speex in our Video
> Conferencing product (see www.sightspeed.com).
--
Jean-Marc Valin, M.Sc.A., ing. jr.
LABORIUS (http://www.gel.usherb.ca/laborius)
Université de Sherbrooke, Québec, Canada
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 190 bytes
Desc: Ceci est une partie de message numériquement signée
Url :
http://lists.xiph.org/pipermail/speex-dev/attachments/20030924/aa6b205a/signature-0001.pgp