Ok, this night I made a test of the different quality-settings in relation to the file-size. I think at least the percentage-values and the average bitrate-value are more or less representative for all produced MPCs.

<Quality-setting> <Avg.Bitrate> <Bitratemin./-max> <Increase in %> <Total Size>
---------------------------------------------------------------------------------------------------------------
--quality5 I 176 kbps I (121 - 214 kbps) I 0,0 % I 908,40 MB
--quality6 I 206 kbps I (156 - 247 kbps) I 17,0 % I 1.067,46 MB
--quality7 I 235 kbps I (187 - 277 kbps) I 33,5 % I 1.215,40 MB
--quality8 I 265 kbps I (213 - 312 kbps) I 50,6 % I 1.371,58 MB

Database: 14 Albums with 171 Songs in Total.
Measured with MPEG-Audio-Collection.

Notes:
--quality 5 = --standard
--quality 6 = --xtreme
--quality 7 = --insane
--quality 8 = --braindead

My opinion to this? - Is it really worth, to save 33,5 % (--insane) or even 50 % more bytes to get a - mostly - nevertheless not audible "better" quality for your files? I publish this, because I noticed, that some guys meanwhile find it "cool" or "fashionably" to encode in --braindead or at least --insane. They say "What do you want, HD-space is cheap in these days", not thinking the aspect, that the files must sent somehow thru the slow internet (with mostly a very small upstream-speed), before they can be stored and played on their PC. In the last few weeks there was even a group founded, which demands at least --insane for their ripps. - I suppose their ripps R.I.P. on their HD´s for even a longer time.... I wouldn´t say a word, if the so generated files would be downloadable from HTTP-servers, so that every broadband-user could use his maximum downstream-bandwith. But it´s not like that....

I know, that I don´t make me many friends in the "high bitrates are the proof for a superb audio-quality"-fraction. But I think a serious discussion has to be lead, as in the other case the things go into a complete wrong direction.

BTW: Once I heard (yet about half a year ago), that --insane (-q7) is not adapted to the human ear. Has this already become outdated?!

- R.A.F. -

This has actually already been discussed many times on the forums here. What it basically boils down to is that people don't like to use --standard because of the title. The basis for their decisions (in at least 90% of the cases) are not because they have found problems with --standard or --xtreme through blind listening tests, but just because they are stubborn and think that somehow these "normal" presets are not good enough (they must all have really special hearing ). Most of the time these people will also completely ignore statements from the authors of these codecs as to whether or not there will be a real improvement in sound quality with these insanely high bitrate presets... even though they'd know better than anyone else.

I think I'll go audition some $6000 power cords for my new amplifier now btw. Got some green markers to try out also!

/me dons asbestos jacket and awaits the hellfire

standard sounds fine to me sure, I have space on my HD for -braindead stuff, but as my music collection increases in size, its sort of unacceptable to use 50% more space, especially when I need space for stuff besides compressed music.

The fact that larger mpc's are harder to share are a moot point for me; I just compress my own music for ease of use, not sharing.

Yes, for sure it is so. It bores me the same like you. But sometimes it´s necessary to warm those well-known topics up. Maybe some of the mentioned people will wake up then (which I honestly said don´t believe)....

There was a time (v0.90 era) when --standard wasn't quite so transparent to some people with some music. However, with the current v1.14 encoder --standard has seen additional tweaks and adjustments since the beta development period. MPC standard files have slowly become larger over time as well. The median bitrate now seems around 180kbps; several encoder versions ago that might have been around 170kbps. While bitrate is not a complete indicator of quality, it does suggest that standard has increased its margin of "safety". For example, the noise-to-masking ratio (--nmt) has increased from 6.0dB to 6.5dB in the past several encoder versions. I also believe Klemm has changed the default ATH profile to a more sensitive model.

I honestly don't think this has had much to do with the people making these claims though. The people in question didn't perform blind listening tests back then just like they don't do them now, and I'd dare say that the quality even on these older encoders was already so high that most people still would not be able to hear the difference.

Standard sounds great to me, so that's what I use. It also saves me about 5-20megs per album compared to LAME -aps. That means I can fit one more album per cd when I make backup copies of my music.

I can understand why people might want to use xtreme just to be on the safe side, but anything above that seems excessive. Especially when some of the sources aren't very complex or hi-quality to begin with.

Most people wud swear that --standard is mostly transparent on 99.99% samples.
And for so long i have been relying on these opinions of HA .

I have to however admit that lately i have been disillusioned after i was listening to my mpc rips on my newly bought audio-system .
I was able to hear artifacts on about 20% of the songs and that makes mpc standard account to 80% transparency .
One such sample can be found here

Still i think it's not good to make --braindead rips .
So i do this - I first make a --standard rip , listen to the album and then if i find artifacts i reencode the wav with a higher setting till the problem stops .And with the above approach i have come to the conclusion that --xtreme owns the 98% transparent tag .
While --insane provides 100% results . B)

what differences are there in the command line and code level tweaking between insane and braindead?

also it should be noted that just because people are not sharing their encodes files via some p2p network doesnt mean that theyll be the only ones to hear them. i always have music playing when im by myself and especially when people are over. people encoding with extreme bitrates maybe doing so not because they believe their hearing to be superior but possibly because they believe their hearing to be inadequate so that others may find fault with their encodes.

I kind of doubt this is why most people use these presets. I've very rarely seen people encode music for other people's hearing. More often than not, you'll see people encode with too low of a setting simply because they don't care about what other people would hear. Even then, if they were doing this, it still wouldn't be very necessary. There are very few people who can discern MPC --standard from the original CD, and even less than can do so with --xtreme. I just can't see why there would be a need to encode with --braindead just because there might be a fraction of the population out there that might run across the file that would need that high of a preset for them to regularly achieve transparency in their encodes.

However, I will admit that it's one thing to encode at such a high bitrate because you are trying reach a point at which all samples, to even people with incredible hearing, are nearly transparent.. but this is really only useful if you are distributing the files to other people explicitly. Not just because you might share them. Most of the people I know who take this kind of approach would be the first to tell you that they don't believe this bitrate is needed to necessarily achieve a very high rate of transparency either... and many of them would also say that they themselves could not hear a difference between the original and a lower preset on the vast majority of samples.

Thanks Dibrom for posting such a thread.

Since those famous *** and **** standard that demand "EAC secure Test & Copy" and "MPC --insane or higher" have been released, i was thinking that HA was becoming a sect rather than a platform that was created to promote high-quality audio to the whole world. I know that HA didn't create those rip standards but they are good examples of what did happened there those last weeks.

U're a great admin, Dibrom. And you're a very great devellopers too, i wished everybody really knew it


edit: please no links ir names of ripping groups.

Ah.. well I didn't actually start the thread (the credit for that should go to R.A.F.), though maybe you're just referring to my comments



Hrmm.. well I'm glad that is cleared up I don't want people to think of HA as a sect or anything like that either, though I know perhaps some people do get the impression that sometimes HA users are a bit too overbearing in their interests for sound quality. What can I say though? I always am and always will be much more interested in measured results than just in speculation and discussion of sound quality based on the "warm fuzzy feeling."



Thanks

I noticed a good example last week : some say that MPC isn't tuned enough to get a good stereo image and want to add --ms 15 to the basic command line.

Now you watch your claims!
Don't mix things up (Even if they are the same in this case - ironically)

"MPC --insane or higher" - quote? from where?
"but they are good examples of what did happened there those last weeks." - Now what does that mean?
Do you have any idea WHY **** uses such high bitrate?
Do you know if we promote this bitrate for general usage?
Hardly.
Keep in mind we're the first to know if anything is modified in Musepack and the first to get the new compiles.

Point being: Don't spit out in such unfriendly manner when you have no clue about the purpose of the project(s).
You are welcome to join the IRC channel some time and ask us how we do our decisions.
xxx.xxxxxxxx.xxx #****
(this is the same network as #Project_Mayhem)


edit: please no links ir names of ripping groups.

The first encoding I made were with Fhg/Radium 128 kb/s HQ : CD quality for several month ; now shit.

Then, I discovered --r3mix : 160-170 kb/s VBR : CD quality, archive setting ; now, too obvious problems.

I switched to --dm-xtreme, then --alt-preset VBR ; standard is great, absolutely transparent during few month. I discovered some classical samples where --preset standard failed, and where I could ABX --extreme.

Really annoyed by the lake of trust I felt with perceptual encoder, I was tempted by lossless. But with more than 1000 CD to encode, it was impossible.

I switched to musepack, and for giving a good margin, I encode my audio-tracks with insane, then extreme, then insane again. I'm not familiar with --standard profil (except for catanets samples ), but when I did ABX test with this profile, I often failed. But with some passage, ABX results were not too bad (11/16 to 13/16). I failed to describe the problem, but not to hear a really small difference - but a real difference...

For that insignifiant reason, I'm happy with my choice. If I, or someone else, will in the next future identify --standard encoding, the choice for --insane gives me a lot of trust. A form of 'quality-insurance'

Listenig music with a perceptual encoder is for me not only a question of pleasure, but a question of safety too. I think that people who encode with --extreme or above are just pessimistic and provident people ; they don't want to regret their choice.

Well, I would agree with anubis in the fact that many people actually encode with --insane or --braindead, without even hearing any difference.
What the point of this? To be in the fashion?

Then yes I bellieve there are difference of quality between the different presets, but the most important thing is really what you hear, not what people think they hear...

I think a simple study would be to take 100 guys from those who are really so sure encoding in --insane or --braindead is better (I bet everybody know some), and make them listen to the same sample in --standard or --extreme.
I bet 90% won't make any difference.

So, what would be great is that each of us would be clever enough to choose his encoding preset because of what he wants & hear, not because "somebody" said it's better

I've really only found one track where MPC's stereo model really messes up: the "White Ladder" track on David Gray's same titled album. I haven't scientifically determined which --ms model provides imaging transparency but the standard --ms 12 is insufficient to the point where I can't overlook it.

I guess I'll be asked to provide a sample.

Perhaps Klemm should just rename --standard to --archive, this would avoid people thinking this quality can't be enough as it's only "standard".

short & perfect !

I consider an 'archive' preset as a transparent setting.
There are problems with --standard (cf. castanets.wav). It's not, for that reason, an --archive setting. Maybe --near_archive

Yeah, renaming standard is a good idea.

If you would rename the preset --standard into --near-archive, guruboolez, then for sure not a single newbie would take this preset anymore.

Maybe it would be good to remove the profile names altogether, and just leave the --quality switch.

Maybe...
I prefer truth than chimera...

A fantastic thing with mpc, is the respect for the user. --radio is very good quality (generally beyond the 'CD-quality' of mp3@128), like a FM radio (so many people assimilate TAPE and FM Radio as shit).
--standard gave a very good sound, but not totally perfect. It's widely 'acceptable'...
--extreme is for the few difficult samples where mpc fail, and for extreme listeners.

This respect differentiate musepack from wma, mp3pro and other propaganda. It's just a different philosophy. I like this kind of 'normativity'.

By calling --standard --archive, newbie will adopt it, but will never believe than small problems can occurs. It will enforce the mythology of the infallibility of --standard profile : not a good thing.
Better enlight newbie than lead them to error.

No, because standard (which is basically archival) is pegged at 5 (on a scale from 0-10). People will not assume that a figure in the exact middle of the quality range provides transparency. They'll say "if 5 is transparent, then why have 6, 7, 8...?" It would add confusion.

Most people feel that you need at least -q 6 to provide best case transparency with Vorbis. With that on their minds, they might not "like" to use --quality 5 with MPC even if their ears confirm there is nothing wrong with this quality level.

Frankly, I wouldn't be in too much of a hurry to change the word presets yet. The worst thing that happens (when people think the word "standard" is too ordinary) is that people encode at too high of a bitrate. Quality-wise, there is no harm done. I guess you could say the same thing would happen if we went strictly with the --quality switch but I can imagine how many will be tempted to use --quality 10.

Well, maybe archive is not a good name, but neither is standard. 'Generally acceptable' quality is mp3 at 128kbps and mpc standard is way above that.

To be honest: I don´t understand the people, who try to archive their original-CD´s into a lossy-codec - even if i´d use highest quality-setting in MPC, it wouldn´t satisfy me at all. Because I would always have the strange feeling: Did I choose the right codec with the right quality-setting? - And maybe later on there´s coming up a codec-revision with better fine-tuning?! Not thinking about the things happen, when a lossy-codec meets an output-source a little bit different from the normal stereo.... "Archiving" means for me, that I can throw away my original CD´s afterwards, and wouldn´t miss anything. And that could never never be, when I did code the music to a lossy format. Even MPC with highest quality-setting --q10 wouldn´t satisfy me. For archival purposes there´s only one choice: LOSSLESS !! Only this does not throw up any questions about quality anymore. So - to come back to the point - even a newbie must know, that the setting "archival" spoken out in context with a lossy codec, means, that they can throw away their originals. So, why not using this expression in the future? - At least this "cheating" would be for a good thing: To push people, who don´t know much about the technical side of lossy codecs, into the right direction and to prevent them from too-high bitrate-encodings.




So it is !

- R.A.F. -

Back when I was using mpc I used standard and was perfectly satisfied with it. I've now moved onto (back to?) lame -aps because I've bought myself a postable. I'm annoyed about the increase in bitrate (which is quite considerable on the kind of music I tend to listen to - mainly metal - which has gone from 170ish to 220ish on average). I briefly tried aac (I have an Expanium, which is one of the few portable players to have aac support), but I could ABX -normal and didn't see the point in going to extreme for only a small bitrate saving over aps (the sample was 'Halo' by Soil, which I couldn't ABX on aps). Also, I don't know if aac will really catch on, and therefore it's quite possible that I might have a portable which doesn't support aac at some point down the line.

Even forgetting about portables, the standard "hard disks are getting so big you couldn't possibly fill them with music, so you may as well go for high bitrate" isn't really valid for me since I store loads of video on HD, so space will be an issue for a long time (especially seeing as video will keep getting bigger for a while: higher resolutions and framerates than DVDs offer do definitely look better - unlike the CD vs DVD-Audio question where this is very much debatable). I don't think I'll have any trouble filling up my hard disk untill they're about 10 TB, which at current rate opf increase will be in about 10 years!

Although I also think that most of the "higher than standard" profile users have not made their choice after extensive ABXing, I disagree about the name issue.
I think the reason that the higher profiles are used is because they're there . Thinking along the way: if Standard would be the ultimate then there would be no extreme and insane.

IMO the names are adequate and well chosen. I mean (e)xtreme, insane and braindead, how much clearer can you get? And standard is in the case of Musepack a high standard, and a good standard it is. It does not say, nor imply inferior or economy mode or moderate as does radio, telephone, wax roll...

Well and then there are those that would like a safety margin for transcoding to other codecs (portables).

Please note that I'm not advocating use of high profiles here, I just think the names are OK.

Have fun ,
--
Ge Someone

.... and yet another good idea, in which name to change the misfitting --standard -preset (even the author hadn´t the intention to give it):
--standard should become --high-standard
That´s now meant seriously, no joke! I think --high-standard - that´s the stone of the wise !!!

But the problem is another, I think: Some people - like those from ***- and **** to say names - think, that it is a kind of "science", to configure EAC right, so that no errors occur while grabbing the CD. To them I must say: Don´t make me laugh! Even an 8 year old average-intelligent school-boy can configure EAC properly (at least after a few hours of trial and error), so that the ripps sound as good as your "perfect quality rips" afterwards. And if there are some drive-configuration-switches, which you are demanded to set or not, and the 8 year old school-boy leaves them as they are pre-set, which role does that play? - As long as the CD is grabbed in secure-mode and the cd-drive is not already 10 years old, everything for sure will also be fine. Even that C2-error-correction isn´t that dangerous as you pretend that it is. Meanwhile I grabbed around 1,500 CD´s - all with having this feature switched on - and I never discovered any errors with it when activated (even on scratched CD´s). And I used not only a single CD-drive, but at least 5 from different manufacturers. Also, switching it off, slows down grabbing-speed of my CD-burner around 2/3 (!). My conclusion: You all see problems, where no problems are (that same goes for the demanded coding-qualities)! - Being a little bit paranoid?! (-> Secure Mode AND Test & Copy?)

- R.A.F. -


P.S.: PROGRAMMING a program like EAC - that´s real science !! THX to André Wiethoff at this place. - But configuring.... bahh!



edit: please no links ir names of ripping groups.

Someone, who had his own R.A.F. Settings (which were pretty messed up) should be carefull when judging tutorials made by others, which just try to achieve max. quality/transparency possible...
dev0

Oh, come on, don't be unfair. I never pretended that this would be "my" settings. Someone else wrote this in the citated thread, as he always read it in my log-files of the albums, which he downloaded from Archmage´s ftp. To be honest, I read about these settings in the former chaostar.org-forum about 1 1/2 years ago and took it, as I did not know better these days. But I don´t use them anymore for more than half a year now. (And please don´t remind me again on my "youth sins". At that time i still was quite young and unexperienced. )
And finally, all has also its good sides: I proofed to be willing to learn, and reacting on criticism - but it seems that others aren´t.

And what do you mean by

Maximum quality/transparency possible? - If you have that goal, you have to go to LOSSLESS - and nothing else. But from the point of bitrate, you are on the best way to go to... No! - What we want - and that is the main-reason, why lossy codecs exist - is very good transparency/quality in relation to file-size! A lossy codec will never be perfect, never mind, which quality-setting is used. Even --braindead for sure has it´s problem-samples

- R.A.F. -