i was wondering what the presets i should archive my cd collection in to acheive the greatest quality. i understand that the extreme, insane, and braindead are the presets that i would be looking at, but i was wondering if someone could tell me the pros and cons of each preset and if these presets have been superseeded by the quality scale that is now being implemented. thank you in advance.

quality 4 - 8 maps to radio - braindead preset. There is no difference. You should first look at standard before you go for anything higher. MPC standard, unlike "standard" WMA @ 64k, was designed to sound transparent with everything you throw at it. There are extremely rare cases where it fails to provide transparency. Most samples that give the standard preset trouble however, won't be transparent at higher bitrates either. So using a higher preset "just in case" doesn't really make sense. You should only take a look at higher presets if you find standard to be insufficient. In this case please report the problem. If you are planning to transcode you might also check out higher presets, but imho if it's just for your portable, then MPC standard will do fine, especially if you use less than lame aps mp3s.

Hi, I'm new at MPC but used LAME --r3mix intensively for archiving my music.

I've tested MPC and out of question, it rules!

Now, I would like to make a collection of Classical music. I've tested some samples coded at --standard and --xtreme, and somehow found the difference to be that in xtreme, I perceive music to be with more room, as if the instruments voices vanished fractions of seconds later and thus the musical experience is felt better.

Now, we come to the obvious question, is all this just perception, or this is in fact a difference? What are other's experiences in encoding classical?

Thanks, Willy.-

Well, you can do some blind comparisons fast with winamp. You will gain high probability easily with ABA (oddman out test).
Just put one encoded (preferably decoded to wav) and 2 originals in to the playlist, randomize and try to catch the encoded 3-4 times in a row (of course randomizing after every guess)

Many people recommend MPC encoder 1.11b (http://www.saunalahti.fi/~cse/mpc/) and the use of --xlevel switch also.

I don't think nobody claims that --standard or --xtreme are totally transparent. You just have to check it yourself using blind testing. --standard is very good for the bitrate though.

Link

So far --xtreme has been very very good to my ears. I cannot hear anything wrong with the streams that I have encoded.

With 1.11b, --standard is excellent for the bitrates it uses (new tests I'm conducting). --xtreme is nearly transparent on most types of music. --insane allocates more bits for those parts of test samples that only the best ears can still hear problems with. Recommended for perfectionists/audiophiles only.

I think the only reason for using higher preset's than xtreme or even standard is the name of the profiles. Most people claim standard won't be enough because it's "only standard".

But MPC was designed to be transparent with the standard setting, and in my opinion it is with 99 per cent of the samples. I was never able to tell any difference so far.

I've noticed this as well. I can pick out --standard encoded files from my playlist without knowing in advance what they are (ie. right after downloading) It's not a real annoying effect, there's no artifacting, it just seems to not sound as good as --xtreme encoded files. It almost seems as if it's not as smooth. Therefore, I personally do not use --standard and insist on rips at least --xtreme. B)

Course, I'm gonna get shot down for saying the above, but I don't care. That's the way it is for me. No I will not ABX, I have nothing to prove, I already know I can hear a difference.

Your own mileage may vary of course.

--xtreme's stereo separation is better on the paper than --standard's. Theoretically this may be the cause of what you are describing. Experiment with --standard and the --ms switch and see if that helps. --standard uses --ms 11 while --xtreme uses --ms 12. Using the combo --standard --ms 12 raises the bitrate by about 3kbp/s when compared to regular --standard.


You make it sound like --xtreme does not sound fully transparent on almost every sample? If I gave you 10 random samples of ordinary music, you can discern original/xtreme on, say, 5 of them (by ear of course)? I find that very hard to believe.

@mpcfiend: what do you expect? You state something that is contradictory to common knowledge and refuse to prove it. I'm not saying you are not telling the truth, but stating something like this without proof is of no value. If you can pick up the difference that easily, it shouldn't be too much of a hassle to run an abx test. Together with a sample this could help to improve Musepack.

I've used ff123's excellent ABC/HR program to test 'noise' music, and spent 7 hours testing 8 different songs (complete ones). 2 samples (one by Merzbow and one by Autechre) were not transparent to me with mppenc 1.11b at --quality 6.5. I could ABX them 10/10. They did sound perfect to me at --quality 7.0. No other claims made as to what other type of music is or is not transparent witn --quality 6.5

Ok, your opinion is valid but it is really uncool of you to:

-Post while knowing in advance that it will annoy the mods
-To say "I won't ABX" this when you know this site mandates blind testing

You can tell people by mouth, over the phone or even your own site, but on HA you will need to ABX before you proceed.

So, yes, you are being shot down. No proof == invalid claim.

Try --standard --ms 12. This uses --xtreme's stereo mode with --standard.

I find it very challenging to believe that you find imaging/stereo problems with --standard but not --xtreme. Standard's default stereo mode may not be perfect but there's just no way you can easily pick between standard and xtreme files based simply on stereo mode. Human ears just don't have that kind of sensitivity.

Unlike some others, I don't feel compelled to beat my chest over my opinions. Everything should be taken with a grain of salt, no matter who it comes from.

Well, nobody is demanding that you have to ABX in order to prove something for yourself, it's just recommended thing to do.
Anyway, ABX can be pretty exhausting, I prefer oddman out (ABA) like I described few messages above.

One thing that sounds pretty strange is that how can mpcfiend "know" from a downloaded file if it's standard or xtreme.. If there's no clear artifacting, you would really need the original for comparison in order to be sure..

If there's somekind of constant quality problem with --standard making it non-transparent pretty much all the time, it could help the developers if mpcfiend could verify this with blind testing...

What are the advantages of 1.11b over 1.1?

Hi, I have a question...

If I use --xtreme --xlevel, will the results be the same as just simply using --xtreme, if the file doesn't have any internal clipping? I didn't use --xlevel with my older encodes (which didn't report any internal clipping), I only started using it recently.

And I must say the quality is very very good! I can't tell the difference between this setting and the original wav.

Yes, there won't be any differences.

No, there's no artifacting. It's quite difficult to find 'artifacts' in an MPC stream to begin with.B) In fact, --standard sounds very good, considering the bitrate. It's more of an impression from the sound of the encoding. It's nothing that I can really put my finger on...strange. It almosts sounds 'fluffy'. Like I said, weird. I don't think ABX'ing would be much help, as the problems go away at -xtreme anyway. So the solution, for me anyway, is --xtreme. Improving the encoder may not be much of an option, since bitrate with a preset is finite to begin with...there's only so much you can do without transferring the problem to another area. 'Fluffy' today, 'stereo image anomalies' tomorrow, so to speak. So, I'll rest easy in the knowledge that I'm in the minority and always use --xtreme or greater. If anyone else finds the same problem, I just want them to know that they're not hearing things, no matter what the vocal majority would have you believe. B)

Anyways, carry on the discussion. I'm done my ranting for tonight.

If they don't trust their own hearing, why should they trust yours? When someone is in doubt whether he hears things or not, the best thing he can do is to perform a blind test (as ABX or ABA).


I wouldn't want to use a codec without tuning, you?

And how can you know if they are hearing things or is it just placebo? You don't know it even in your case, since you haven't really tested it in order to have any statistical significance to your claims..

I'm not at all saying that --standard is transparent if the choises are "transparent" or "non-transparent", and I think the vocal majority knows it.
The thing is, there are so many claims about MPC (and all the other codecs) quality problems, that if all were completely accurate, you'd think the quality would be like Xing-96kbps. As always, most of the time these people don't bother showing any evidence, because they "know" it. Most of the time they just disappear. Few times I've seen people doing test when asked and after testing, the "absolutely certain" difference wasn't there after all. Few times people with problem reports have ABXed succesfully and provided even samples. There are examples in this forum about those cases.

Fact is that MPC is not totally transparent always, and it's perfectly possible to ABX/ABA --standard, but does it have problems you describe? The problem arises if you insist here that there's a problem you describe, but refuse to prove it even though obviously you have not even proved to yourself if what you think is true or not.

Does this all matter? Well, it would be nice if (even personal) recommendations would be based on even somekind of provable result(s), not totally on unproven "certain feelings" which so many times have turned out to be impossible to hear in blind test. Of course imo it's perfectly ok, if you say that you want to use use --braindead --nmt 55 --tmn 55 just to feel safe, or because you just like to use it...
But if you say you use that, because you hear problems with --nmt 54, then it would be nice to know that you have actually proved it to yourself that you really hear a difference.

Hmm, what I think I want to say with all this is that, blind testing is really encouraged and recommended here, but I think you knew that already..
It's also recommended in order to provide facts for developers in order to help the tweaking.

Just a note for those reading mpcfiend's reports of differences between --standard & --xtreme - Ingore it.
Don't jump on his train until You, can hear this for Yourself.
Statements like this, without Proof, should always be ignored. Censored IMO