Originally posted by windoze9x
I've been archiving albums to mp3 and I want to keep them in archival quality while being able to archive 1000+ cds (so no lossless).  I have been very happy with the --remix preset using razorlame.  Alot of people say this is outdated and I should be using --alt preset standard.  Is this true??

I know that r3mix.net had a lot of research to back up their settings.  HMM...I want to keep around the same filesize.

Does ANYONE think r3mix is better than alt standard?

Originally posted by Bluenote
Why do you guys find it so hard to accept some people out there might actually understand the difference between --r3mix and --alt-preset standard, yet still use --r3mix at times?

I am one of these people. I do not doubt that --alt-preset standard is the qualitatively better standard, and I take my hats off to all the guys who spent to much time implementing it. At the same time, I have never been able to hear any defects in --r3mix encoded files (blame it on my decidedly non-golden ears ) and the one thing you guys keep claiming is that the bitrate difference is only "a few kb/s on average". That is simply a lie. I find that on average (using charts-type pop music) --alt-preset standard is roughly 20kb/s more, and that is something significant. You need to admit this, guys, and you also need to admit that Roel never claimed his settings were perfect, just that they represented his personal compromise between quality and file size.

And since using lossy audio compression will always involve some kind of a tradeoff, I think this is perfectly valid. Those that want the state of the art, can use --alt-preset standard, as I do myself plenty of times. Those with smaller hard disks sometimes also use --r3mix.
I really don't wish to start a flame war here, but I do get the feeling --r3mix has been unfairly represented here at times.

Originally posted by windoze9x
hey guys...I don't want to cause a commotion but I just wanted to know which is ACTUALLY better.  I mean they really do produce a similar filesize.  I have encoded my mp3s so far in --remix setting and am not going to re-rip/encode them.  but I wanted to know if it's worth the switch.  Which apparently it is.  Thanks for the suggestions.  thanks ALOT actually

Originally posted by William
I hope that we can seperate the --r3mix preset and Roel's attitude towards quality improvement.

1) Before the invention of alt-presets (and the older dm-presets), was there anyone not satisfied with --r3mix?

2) Is --r3mix better than normal -V options?

3) Is --r3mix worse now than in the past?

4) Any presets that gives better quality at the bitrate of --r3mix?

IMO, we get what we pay for. With enough testing, code-level tweaking, and at about 20 - 30 kbps larger, alt-preset standard must be better than --r3mix. But can it replace --r3mix? I doubt it, especially for those with bitrate concerns.

I think the presets focus on different "markets". For archival quality, alt-presets are the one to be used. However, at lower bitrates and still good quality, --r3mix still has its life.

    --alt-preset type type must be "standard", "extreme", "insane",

                      or a value for an average desired bitrate and depending on

                      the value specified, appropriate quality settings will be used.

    --r3mix         use high-quality VBR preset

LAME version 3.92 MMX  ([url]http://www.mp3dev.org/[/url])



Presets are shortcuts for common or carefully tuned settings.

Several separate collections of preset profiles are available.

For more preset details, refer to --preset longhelp.



basic- presets covering a wide span (fixed rates unless used with -v or VBR):

    phone (16 kbps)    mw-us (40 kbps)    radio (128 kbps)    studio (256 kbps)

    phon+ (24 kbps)    sw    (24 kbps)    tape  (128 kbps)

    lw    (24 kbps)    fm    (112 kbps)    hifi  (160 kbps)

    mw-eu (24 kbps)    voice (56 kbps)    cd    (192 kbps)



alt- presets highly tuned for utmost quality via blind listening tests:

  VBR presets for steady quality

    --alt-preset standard

    --alt-preset extreme

    --alt-preset insane

  ABR presets for best quality at a given average bitrate:

    --alt-preset <bitrate value>



r3mix- VBR preset for steady quality with little excess:

    --preset r3mix

[size=9]

C:Downloadslamelame-3.92>lame --alt-preset help

LAME version 3.92 MMX  ([url]http://www.mp3dev.org/[/url])



The --alt-preset switches are designed to provide the highest possible quality.



They have for the most part been subject to and tuned via rigorous double blind

listening tests to verify and achieve this objective.



These are continually updated to coincide with the latest developments that

occur and as a result should provide you with nearly the best quality

currently possible from LAME.



To activate these presets:



   For VBR modes (generally highest quality):



     "--alt-preset standard" This preset should generally be transparent

                             to most people on most music and is already

                             quite high in quality.



     "--alt-preset extreme" If you have extremely good hearing and similar

                             equipment, this preset will generally provide

                             slightly higher quality than the "standard"

                             mode.



   For CBR 320kbps (highest quality possible from the --alt-preset switches):



     "--alt-preset insane"  This preset will usually be overkill for most

                             people and most situations, but if you must

                             have the absolute highest quality with no

                             regard to filesize, this is the way to go.



   For ABR modes (high quality per given bitrate but not as high as VBR):



     "--alt-preset <kbps>"  Using this preset will usually give you good

                             quality at a specified bitrate. Depending on the

                             bitrate entered, this preset will determine the

                             optimal settings for that particular situation.

                             While this approach works, it is not nearly as

                             flexible as VBR, and usually will not attain the

                             same level of quality as VBR at higher bitrates.



The following options are also available for the corresponding profiles:



   <fast>        standard

   <fast>        extreme

                 insane

          <cbr> (ABR Mode) - The ABR Mode is implied. To use it,

                             simply specify a bitrate. For example:

                             "--alt-preset 185" activates this

                             preset and uses 185 as an average kbps.



   "fast" - Enables the new fast VBR for a particular profile. The

            disadvantage to the speed switch is that often times the

            bitrate will be slightly higher than with the normal mode

            and quality may be slightly lower also.



   "cbr"  - If you use the ABR mode (read above) with a significant

            bitrate such as 80, 96, 112, 128, 160, 192, 224, 256, 320,

            you can use the "cbr" option to force CBR mode encoding

            instead of the standard abr mode. ABR does provide higher

            quality but CBR may be useful in situations such as when

            streaming an mp3 over the internet may be important.



    For example:



 or "lame --alt-preset fast standard <input file> <output file>"

 or "lame --alt-preset cbr 192 <input file> <output file>"

 or "lame --alt-preset 172 <input file> <output file>"

 or "lame --alt-preset extreme <input file> <output file>"

Originally posted by Dibrom

Yes, quite a few.  HA was eventually formed because of this actually, because the discussions for trying to provide a better presets were no longer possible on --r3mix.net.

Almost all of the prominent members of this community were people who originally helped with --alt-preset development in some way or another (usually testing and discussion), and almost all of them did so because of the problems they had found with --r3mix.

You'll have to be more specific than this.

I've begun to suspect that the newer --r3mix is actually worse than the old --r3mix in more situations than not actually.  The newer --r3mix switched to using nspsytune, but then skimped out in a few other areas, which I believe probably led to much worse pre-echo handling in the majority of cases, even if a handful of more widely known test cases were improved.  Of course, both are non-optimal compared to the --alt-presets, so I haven't gone much out of my way to verify this theory.

--alt-preset standard -Y

I'm also kind of working on another preset at the moment that will be closer to/lower than --r3mix in bitrate, but that's on hold right now while I work on the FAQ's and new website.

There are better alternatives to --r3mix.  --alt-preset standard -Y for example.

That's funny, I thought it already had replaced it.[/b]

Originally posted by Destroid
So, yeah, I guess to show appreciation there's a notable mention of the site that helped LAME become popular. But it should really be mentioned in the docs or the credits (i.e. LAME version 3.92 MMX  (http://www.mp3dev.org/    For more tests/info visit www.r3mix.net).

Originally posted by William

lame --r3mix test.wav
lame -V0 test.wav

Which one is better?

I am looking forward to this new preset.
Are there any samples that shows --alt-preset standard -Y to be superior than --r3mix?
If --alt-preset standard -Y is better, then everyone can happily switch to it.

For archival quality, --alt-preset standard has replaced --r3mix. That's no doubt.
But in the case of portable MP3 players with limited memory, --r3mix might still be the choice because the MP3 player can hold more songs and for causal listening, it may not be that bad. Of course, --alt-preset standard -Y can also be the choice in this case if it is better.

I would like to see some comparison between --r3mix and --alt-preset standard -Y.

Originally posted by Destroid
But it should really be mentioned in the docs or the credits (i.e. LAME version 3.92 MMX  (http://www.mp3dev.org/    For more tests/info visit www.r3mix.net).

Originally posted by JohnV
God help us, I don't want to see those comments appear even more often... IMO r3mix is actively hurting the community by spreading false info and weird impressions. Even developers of other formats don't like it, because it spreads totally false impressions of how codecs should be tested.

Originally posted by fewtch
To be fair, he does include appropriate disclaimers at the top of the "analysis" page:

"Tests and fancy graphs don't prove much.  Anyone with a bit of computing experience can look for some piece of music, and produce an "encoder comparison" so that their favorite mp3-er comes out best."

Originally posted by Destroid
The politics you're referring to must be the r3mix site helping show the flaws of other encoders.

Originally posted by JohnV
It doesn't show flaws of other encoders. It just shows some frequency graphs. How much those make difference to the actual perceptual audio, is another thing. You don't look flaws of perceptual lossy encoders by looking at the FFT graphs. Or using totally tonal sinesweeps or totally noisy white noise, what does that possibly could tell about the true quality of a psychoacoustic-model for example??