[vorbis] RC2 artifacts

[Gian-Carlo Pascutto]

----- Original Message -----
From: "Bacchus 13" <bacchus13 at zdnetonebox.com>
To: <vorbis at xiph.org>
Sent: Thursday, August 16, 2001 9:58 PM
Subject: Re: [vorbis] RC2 artifacts

> ...Encoding doesn't seem to be so much different since beta 4 (By the
> way, did you update the decoder as well if you are on Windows?  For there
> seem to have been a bug-fix in RC2).  Actually, I expected even worse
> results by the added channel coupling.

Worse? If anything, channel coupling should drastically improve
quality at equal bitrates. Channel coupling removes more redundant
information that can be used to improve the hard to encode parts
of the music.

The only problem that it has now is that a piece of normalisation
code is missing that can add a hissyness at bitrates < 256kbps if
point/phase stereo is used.

> However, the new lower bitrates options seem to be better than its rivals.
>  I don't know why they bother comparing WMA at 64 with MP3 at 128 in r3mix
> forum (http://r3mix.net/forum).  Obviously, Ogg sounds much better and
> is smaller than WMA.

What are you talking about? The listening test going on in r3mix is
comparing WMA,OGG,LAME,Xing,AAC and MPC, all at 128kbps (or VBR setting
which averages that)

If people here are interested in participating, please take a look at
http://66.96.216.160/cgi-bin/YaBB.pl?board=c&action=display&num=997771687

This test is still is testing phase :)  The first sample is used to
gather comments about how to improve the test, and to make people familiar
with the testing methodology. The actual test will start when RC3 comes out.

If you want to help gather more reliable information about how Vorbis
compares to other codecs, please join the test. This should be excellent
data to counter Washington Post-like FUD.

> > I noticed some in beta4 and was hoping they would disappear with RC1.
> > However, I've been playing with RC2 and I've still been banging my
> > head into a number of problems.

I did not like some aspects of the RC2 encoder and have my own tuned mode.
It fixes the problems that were most annoying to me. It uses about
150-180kbps
depending on the exact music content.

prebuilt Win32 binaries with this tuning are available at

ftp://sjeng.sourceforge.net/pub/sjeng/vorbis/tuned (retry if it fails to
connect)

Sources are in the src/ dir, but you need to integrate the new mode into
libvorbis yourself (good luck ;)

Enabling it in oggdrop should be obvious, for oggenc you just need to
specify
a really large bitrate (e.g. -b 999)

I tested it with your sample and it seems to fix the problem to me.


--
GCP

--- >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-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.