Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: --r3mix 3.95 vs. 3.93, bigger files? (Read 12019 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

--r3mix 3.95 vs. 3.93, bigger files?

Reply #25 –
Quote
Flash forward an eon or so. Several capable individuals, including some of those directly involved with coding LAME, put forth a massive amount of effort to create a transparent-all-the-time setting for LAME (within the limits of LAME and the MP3 spec to do this). They were focused on dealing with LAME 'gotchas' that the --r3mix creator was not aware of or did not account for, and utilized verifiable public testing to determine on a larger scale the threshold of 'transparent' to the general audience.

Do you any more information concerning this public test that supposedly occurred?  I've heard a number of people mention this, but I don't recall seeing any.  I'm sure it happened, but my skeptic side would like to see evidence of this.

I also ask because the next generation codecs are in dire need of this.  Vorbis needs to be tuned better, preferably with public testing.  The same might be said with FAAC.

Can you provide anymore information on the --aps public testing that occurred?

--r3mix 3.95 vs. 3.93, bigger files?

Reply #26 –
There was a major test carried out by r3mix himself comparing r3mix with the (then) dm presets. Unfortunately the results of this test were hosted on the r3mix website, which is now dead. I have copies somewhere, if you're really interested.

The subsequent development of the dm presets into the alt presets is documented on this board - though it's easier to say this than to find it! Have a search through the archives if you're interested.

Cheers,
David.

--r3mix 3.95 vs. 3.93, bigger files?

Reply #27 –
Not to be contradictory or anything but I have seen the tests david speaks of they are around the web on some obscure mirror of the r3mix board. If you desire it enough it can be found.
r3mix zealot.

--r3mix 3.95 vs. 3.93, bigger files?

Reply #28 –
Quote
Do you any more information concerning this public test that supposedly occurred?  I've heard a number of people mention this, but I don't recall seeing any.  I'm sure it happened, but my skeptic side would like to see evidence of this.

From what I remember, the development of the 'presets' was basically a matter of volunteerism reaching critical mass, like any active open source project. Individuals would contribute discovered problem samples, and Dibrom and company would alter the presets (or LAME itself) to better handle them. Not really structured per se, but it still allowed active outside participation, and the ability for decisions to be questioned.

In comparison, --r3mix development involved one individual, charting out seemingly optimal off-the-shelf settings based on studies of human hearing. This was a vaild approach, but it assumed that LAME (and MP3) worked optimally and as expected.
Quote
I also ask because the next generation codecs are in dire need of this.  Vorbis needs to be tuned better, preferably with public testing.  The same might be said with FAAC.

The only thing really keeping this form of development from occuring is lack of a 'leader' with sufficient ability and time to make it happen. As the membership of this board shows, there are more than enough problem sample submitters. Sadly, as witnessed with the 'presets' and Garf's GT vorbis, this appears to be a thankless and unpaying job. Perhaps some manner of incentive is in order, ala the PC for Frank Klemm.

Quote
BTW I doubt many people here slinging at --r3mix could ABX the difference to Darin's presets on their music.

The same could easily be said for 192kbps CBR Stereo MP3's, until things go wrong...

--r3mix 3.95 vs. 3.93, bigger files?

Reply #29 –
Quote
Not to be contradictory or anything but I have seen the tests david speaks of they are around the web on some obscure mirror of the r3mix board. If you desire it enough it can be found.

If you know where this is, please share! Google isn't my friend on this one!


I've found the r3mix results quoted in my thesis - it's difficult to re-format them to post here, but...

The tested encoders / settings are:
lame 3.90a7 and MPEG+ (unknown version)

1. lame cbr192 MS GPSYCHO (192kbp)
2. lame abr224 MJ nspsytune nssafejoint (221.3kbps)
3. lame --r3mix (199.6kbps)
4. lame dm-xtr (231.8kbps)
5. MPEG+ insane -nmt99 -tns99 lowpass 19.5 (near lossless achor 740kbps)
6. lame dm-ins (272.8kbps)
7. lame cbr256 nspsytune nssafejoint (256kbps)
8. lame dm-std (221.9kbps)

Code: [Select]
codec:     1    2    3    4    5    6    7   8
result 01 4.0 4.5 4.1 4.5 5.0 4.0 4.0 4.3
result 02 3.7 3.9 4.0 4.2 3.8 3.5 3.2 3.7
result 03 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 04 4.0 4.0 4.0 4.0 5.0 5.0 5.0 4.0
result 05 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 06 4.5 4.7 4.7 4.7 5.0 4.7 4.7 4.7
result 07 3.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 08 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 09 3.3 4.0 3.3 3.0 5.0 3.2 3.8 3.5
result 10 4.0 4.4 4.5 4.2 4.7 4.3 4.6 4.1
result 11 3.7 4.0 4.5 4.5 4.5 3.7 3.7 5.0
result 12 4.0 4.0 4.0 4.0 4.0 4.0 4.0 4.0
result 13 4.4 5.0 4.5 5.0 4.9 5.0 4.6 5.0
result 14 4.0 3.5 3.0 3.5 4.0 3.5 3.5 3.5
result 15 2.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 16 3.0 2.0 2.5 3.0 4.0 4.0 3.5 5.0
result 17 3.5 4.0 4.7 5.0 5.0 5.0 4.9 5.0
result 18 3.5 4.0 3.5 5.0 4.0 3.0 3.0 4.0
result 19 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 20 4.0 4.2 4.3 4.7 4.6 4.1 4.4 4.5
result 21 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 22 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 23 4.2 3.8 3.8 4.0 4.0 4.5 4.3 4.2
result 24 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 25 4.8 4.8 4.8 4.8 4.8 5.0 4.8 4.8
result 26 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 27 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 28 5.0 4.9 5.0 5.0 4.9 5.0 5.0 4.9
result 29 4.9 5.0 4.9 4.9 4.9 5.0 5.0 4.9
result 30 4.0 4.0 4.0 4.0 5.0 5.0 4.0 5.0
result 31 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 32 4.6 4.8 4.6 4.6 4.4 4.6 4.6 4.6
result 33 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 34 4.5 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 35 4.5 4.2 5.0 5.0 5.0 4.4 4.2 5.0
result 36 5.0 4.0 4.0 4.0 5.0 4.0 4.0 4.0
result 37 4.3 5.0 4.8 4.8 5.0 4.7 5.0 4.7
result 38 4.9 4.5 4.6 5.0 4.9 4.9 4.9 4.5
result 39 5.0 4.8 4.6 5.0 5.0 4.5 4.5 4.7
result 40 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0
result 41 3.4 4.2 3.5 4.4 3.9 3.7 4.0 4.5
result 42 5.0 5.0 5.0 5.0 5.0 5.0 5.0 5.0


Scores, in rank order (out of 210):
200.3 MPEG+ (high anchor)
196.1 dm-std
194.6 dm-xtr
192.3 dm-ins
191.2 cbr256
190.4 abr224
189.2 --r3mix
183.7 cbr192

This is a very old test, conducted using an older encoder version, and older versions of the presets
This test was not conducted using all the safeguards which are normally employed here
(It would unfair to imply harsh criticism, because it was one of the first net audio tests carried out, and probably formed part of the learning process for many others)

Useful things to note:

a) look how many people ranked everything 5.0 (or 4.0), and that others misranked the (exceptionally) high anchor. (There was no ABXing required for this test, though some people did used ABX)

b) r3mix published the results of this test, even though it suggested that his preset command line was inferior.

IIRC All presets were changed after this test

Cheers,
David.

--r3mix 3.95 vs. 3.93, bigger files?

Reply #30 –
Quote
If you know where this is, please share! Google isn't my friend on this one!

Google will be your frient once it will have properly indexed your thesis. (please, next time you write a master thesis, please try to fit it under 2MB and not 20MB)

--r3mix 3.95 vs. 3.93, bigger files?

Reply #31 –
Here is one mirror that I just Google'd up:
r3mix forum mirror

--r3mix 3.95 vs. 3.93, bigger files?

Reply #32 –
Quote
Here is one mirror that I just Google'd up:
r3mix forum mirror

Is it just me, or did the community at r3mix back then seem a lot smaller than the one at HA?

For comparison, links to the first 5 topics posted at HA (at least, based on URL):
http://www.hydrogenaudio.org/forums/index.php?showtopic=1
http://www.hydrogenaudio.org/forums/index.php?showtopic=2
http://www.hydrogenaudio.org/forums/index.php?showtopic=3
http://www.hydrogenaudio.org/forums/index.php?showtopic=4
http://www.hydrogenaudio.org/forums/index.php?showtopic=5

EDIT: added links

--r3mix 3.95 vs. 3.93, bigger files?

Reply #33 –
Quote
Here is one mirror that I just Google'd up:
r3mix forum mirror

That's not a mirror. That was the first version of the board which ran for a few months using EZboard hosting.

EZboard added more and more adverts, so the forum moved to private hosting. It's the huge privately hosted forum which has vanished. It was dead anyway in the end, but the archive of posts was very valuable. 

Cheers,
David.

--r3mix 3.95 vs. 3.93, bigger files?

Reply #34 –
Quote
That's not a mirror. That was the first version of the board which ran for a few months using EZboard hosting...

Very true, I did not look at it for very long, and it bit me for the inattention. The forum did seem extremely incomplete from what I remembered of it (including it's last days), but I figured it was due to bit rot or incomplete transfer. Now I know why. Thanks.