[vorbis-dev] video codec

[Keith Wright]

> From: Steve Underwood <steveu at coppice.org>
> 
> Marco Al wrote:
> > 
> > That's not the point, its the user's responsibility to present the
> > images as he would like them compressed. Suggesting using a noise
> > polluted source to test is doing things backwards... if it removes
> > them its ok and if it doesn't remove them its still ok, well that
> > was a usefull test.  Test with an image sequency which naturally
> > has high frequency components.

Yes, do that.  But also test with noisy source to be sure the
compressor does not make the noise worse.

> What you say is really no answer. The user seldom has control. If I
> have a noisy movie to compress, that is what I have to
> compress. Clean it up?  Well, I can filter a bit. If I filter a lot,
> though, I'll loose lots of detail.

A good argument _not_ to make filtering a part of the compressor.
Have many optional filters _before_ compression, some may make
the image look better, some may make it compress better, but the
goal of the compressor must be output that looks just like the
input.  Noisy in, noisy out.  But not _more_ noisy.

> A related case. Digital cell phones sound OK in a quiet room, but sound
> pretty awful out in the street where there is background noise. The
> nature of voice compression makes this difficult to avoid. However, in
> the competitive processes to choice the winning vocoder 100% of all
> testing is conducted in a studio environment. Not a since utterance
> other than a single clean human voice has been used for the competitive
> evaluation of the vocoders we use today.
>      <the engineers did a shoddy job, 
>         because the specs ignored noise>
> I think this sucks. I was amazed when I was first introduced to this
> insanity. Test hard with the problem cases, but keep a close eye on them
> all.

To the extent this is true, it is sucky.  But I have a document called
"Draft Specifications for the A/D Conversion of Voice by 2400b/s Mixed
Excitation Linear Prediction".  (I believe this is now a US government
standard, it was certainly intended to be.)

On page 17 it says:
> The Diagnostic Rhyme Test will be performed in accordance
> with ANSI standard S3.2-1989...
Then there is a table called "Weights and Thesholds for
Intelligibility Conditions".  Among the conditions we find
"Quiet", "Office", "MCE Field Shelter", "Car", "M2 Bradley",
and "F-15".  So at least some standards require testing
under noisy conditions.  If this was not done for cell
phones, perhaps it is because nobody would be so foolish
as to attempt to talk on the phone while driving. (right?) 


-- 
     -- Keith Wright  <kwright at free-comp-shop.com>

Programmer in Chief, Free Computer Shop <http://www.free-comp-shop.com>
         ---  Food, Shelter, Source code.  ---

--- >8 ----
List archives:  http://www.xiph.org/archives/
Ogg project homepage: http://www.xiph.org/ogg/
To unsubscribe from this list, send a message to 'vorbis-dev-request at xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is needed.
Unsubscribe messages sent to the list will be ignored/filtered.