Vorbis - Jan 2002 - Interactive RC3 quality analysis graphs

This is the new page after the first feedback round:

http://audio.sinderman.com/

Comments? Conclusions? Developers feedback?

Can someone ABX 60.wav with 60.ogg -q 0?

Cheers, AGS.

P.S. Garf/Erik, I have included some comments from EAQUAL's author.

<p>--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

> Can someone ABX 60.wav with 60.ogg -q 0?
Yes, it is not transparent.  It sounds like tape flutter.  Here are my 
results, all reliably ABX'd to a p-value of at least 0.01:

-q 0, ABX 10/10, bitrate = 22.2 kbit/s
-q 4, ABX 8/8, bitrate = 26.4 kbit/s
-q 5, ABX 16/19, bitrate = 27.5 kbit/s
-q 6, ABX 9/10, bitrate = 34.1 kbit/s
-q 9, ABX 8/8, bitrate = 43.4 kbit/s

It does get better with increasing quality, but by the time I got to -q 9, 
I had trained myself to hear it well enough.

ff123

<p>--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

On Sunday 20 January 2002 17:00, AGS wrote:
> This is the new page after the first feedback round:
>
> http://audio.sinderman.com/
>
> Comments? Conclusions? Developers feedback?
>
> Can someone ABX 60.wav with 60.ogg -q 0?
Yes. The high tone warbles/flutters at -q 0. It doesn't seem to at -q 3, so
I
assume it's mainly because those levels haven't been fine tuned yet.

John

--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.

When I first tried RC3, I tried -q 0 on a somewhat mellow acoustic track
(encoded to 56 kbps), and was amazed - although you could tell it wasn't CD
quality there weren't any harshly annoying artifacts. I then tried a busy
hard rock track, and the results were drastically different (pretty
distorted). 

So it seems the results are highly dependant on the music you encoding at
quality level 0 (I'm guessing you encoded classical or something acoustic?).

I've encoded and listened to about 15 of my CD's at -q 3 now, and
I'm very
impressed. Although I haven't A/B'd original vs -q 3 Vorbis, I
haven't been
able to pick anything offensive out (like the crappy cymbals you can get
with lower bit rate MP3). Maybe it's just a matter of not being tuned to the
type of artifacts Vorbis produces (took me a while to realize how bad MP3
was initially), but hopefully not. The bit rate averages about 110 kbps at
-q 3. The only possible bad thing I've noticed is the high frequencies
don't
seem to be very present in the scope, but my ears can't casually discern a
quality loss with an A/B.

Anyway, RC3 seems to be rock, I'm very impressed!

Thanks,
Dave.

-----Original Message-----
From: Kenneth Arnold [mailto:ken@arnoldnet.net]
Sent: Tuesday, January 22, 2002 5:42 PM
To: vorbis@xiph.org
Subject: Re: [vorbis] Interactive RC3 quality analysis graphs

<p>On Tue, Jan 22, 2002 at 08:19:09PM -0500, Monty
wrote:
> People are seriously bothering to ABX/expect any kind of quality from
> -q 0?  The 'fuck it, there's no hope, just chop bits off' mode?
Eek.
You mean the virtually transparent to the unskilled / casual listener
mode? I consider myself to have better than average ears, yet I really
have to listen carefully back and forth to hear the difference between
-q0 and original. -q0 Vorbis seems a little duller on the track I just
tried, but it's barely noticable, and for those who consider WMA at
48k "near-CD quality", I doubt most people would really care.

<p>--- >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@xiph.org'
containing only the word 'unsubscribe' in the body.  No subject is
needed.
Unsubscribe messages sent to the list will be ignored/filtered.