[speex-dev] Way to measure loss of quality

Jean-Marc Valin jean-marc.valin at hermes.usherb.ca
Wed Aug 20 00:32:15 PDT 2003

> I find it difficult to hear 2 voice samples and tell
> which is nearer the original, especially if the
> background hiss is slightly different. So what if you
> actually subtract the post-compression sound from the
> original and then listen to the DIFFERENCE. If you
> can't hear any voice except background noise and some
> hiss from "s"'s and "d"'s and such, that means most of
> the actual voice has been maintained. If you can hear
> someone speaking, that means there is leftover voice.
> If you can actually understand what that someone is
> saying, the difference is great. Thoughts?

What do you prefer: a codec that sounds good or one for which
subtracting the original gives a better signal? 

> My hearing is not the best I suppose, because I can
> barely tell the difference between any of the
> encodings and the original for the examples hosted on
> the Speex.org website. When I encode, I can barely
> tell the difference between quality 3 and quality 10.
> Complexity is at 8 or above. SO my questions are:
> 1. What would be a common quality level for 56k modems
> to use in voice chat during a game? CPU usage is not
> an issue.

For 56k, you have ~36kbps upload. If you remove the UDP/RTP overhead,
that leaves you about 10 kbps, so I'd suggest the 8 kbps mode. The 11
kbps would probably also work.

> 2. Would it be better to have 16khz at quality
> 3(complex10) or 8khz at quality 5, they end up roughly
> the same size? (let me know if you are refering to VBR
> or not or both)

Well, what do *you* like better. At this bit-rate, it's mostly a matter
of taste, though at higher bit-rate you'll probably want to go with 16


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/20030820/3a5e84d4/signature.pgp

More information about the Speex-dev mailing list