[vorbis] RC3?

[Greg Wooledge]

> > >> 1 / Voice Grade
> > >> 2 /
> > >> 3 / FM radio

> > > That won't work since Vorbis is simply better than that. Quality 1 to 3
> > > is already what the common public regards as "near CD quality".

> > I disagree and think EdB's scale is accurate. Maybe I'd count 6 as
> > audiophile, but I can hear the difference between 5 and 6.

(etc.)

What we need are real statistics.  Then we can say things like "-q Y
has a percentile rating of X%" which would mean "X% of listeners cannot
distinguish -q Y from the original on the following reference samples
(blah, blah) with the following hardware (blah, blah)."  This would
give us a rating that would at least be comprehensible to people who
are willing to think, and which is not entirely subjective.

Of course, this would require a massive set of data (listening tests
with real hardware), and therefore time.

Meanwhile, I'd say just use the -q value and the nominal bitrate that
this -q value produces on typical songs (3 => 112, 4 => 128, etc.).
Anything else is meaningless.


-- 
Greg Wooledge                  |   "Truth belongs to everybody."
greg at wooledge.org              |    - The Red Hot Chili Peppers
http://wooledge.org/~greg/     |


-------------- next part --------------
A non-text attachment was scrubbed...
Name: part
Type: application/pgp-signature
Size: 241 bytes
Desc: not available
Url : http://lists.xiph.org/pipermail/vorbis/attachments/20020103/abb55d2b/part.pgp