*** NOTICE ***

--quality and --xlevel are safe to use.
--xlevel is needed to prevent clipping during encoding
(this won't be necessary anymore in the upcoming SV8 encoders).


The answer was written in first post.
(well, I admit, that previously the combined information was spreaded in 2 parts..., now edited.)

I noticed on Case's Musepack Page that there is a 1.15j alpha version available for download. Is there a reason why this isn't recommended for use?

You can't stay updated all the time
1.15j is just some bugfix release after 1.15f.
it's "safe" to use and has sofar proved to perform best of the available SV7 encoders.
edit, yeah, alpha is alpha :B

Yes, it's because we don't recommend alpha versions. Even though there may be slight advantages in quality and/or usability, these are generally untested and may be unreliable. Many times, the quality improvements are not big enough to justify a recommendation of an untested version. Also, in retrospect, the changes in an alpha version may turn out to be flawed (although i agree that this doesn't often happen). Lastly, alpha versions generate the profile tag "unstable/experimental", which renders the profile info useless.

2003-04-10 Added:

CDIndexer : for CD-Audio & data-CDs
MAC 2.90 : Mpeg Audio Collection 2.90 : perfect for mpeg, not yet for CD-Audio



Should we change something at list, regarding 24 bit dithering, as sugegsted in that mail/post above ?

newbie question

should i use both -extreme AND --xlevel switch?

when will the use of the -xlevel switch be eliminated?
where can i find detailed info and guidelines regarding HQ mpc encoding, and what every switch actually does?
what is SV 7?


thank you

@alfa156:

All you've asked for is in the beginning of this very thread...

And we've got Search function here also...

newbie question

should i use both -extreme AND --xlevel switch?

extreme etc are names for the quality switches,

use --quality 5, 6, 7 or 8 plus --xlevel !!!. (6.00 - 6.99 is/was called "extreme")


when will the use of the -xlevel switch be eliminated?

When it is written there in first post of this thread, then you don't need to use it anymore with according version

where can i find detailed info and guidelines regarding HQ mpc encoding, and what every switch actually does?

Guidelines for HQ encoding:
erm, of course in the first post here...
or *

what every switch does: have a look in first post, there are the there recommnded switches explained. More: here in forums, mpc technical or general.


what is SV 7?

StreamVersion 7 of mpc. SV8 is worked on.


thank you

You are welcome !


MOD: *please no links to ripping group guides.

edited by JanS 18th April, 2004

strange, that nobody else complained about group more than 1 year.
users question:
which group ?

24-bit capable does not mean 24-bit accurate.

In this world, it is impossible to make a "perfect" 24-bit ADC or DAC - you can't make a device that has absolutely no noise down to this level. I think that's what the FAQ is saying.

However, there are some convertors that are 24-bit "accurate" - this means each value it outputs is accurate, on average. In other words, the response of the device is "linear" down to some level. Early 16-bit DACs weren't even linear to 16-bits - after averaging away any noise, some of the smaller numbers wouldn't even come out in the right order! You put in 1,2,3,4,5 and get out 1.2, 2.9, 2.6, 3.8, 5.1 for example!

In a device that is linear down to 24-bits, the values are swamped by noise, but if you average that noise away (a process that the human ear is capable of), what's beneath it is correct. Very few convertors are 24-bit accurate by this description, but some are. There's a discussion of this topic here:
http://www.dcsltd.co.uk/papers/bits.pdf
which refers to one of their own products, the DCS Elgar:
http://www.dcsltd.co.uk/elgar.htm
...which costs more than a reasonable car.

Cheers,
David.

thank you for your quick answer!

Of course i have looked the whole thread (not just the first post) before posting, and yes, i know that there is a search function in this forum. Regarding the switches, i wanted to know more technical details, and that's why i posted. I don't think that this is just my impression but when i started to mess with HQ encoding it was much easier to find (detailed) info about the mp3 format than about mpc. Than's why i asked for some help, and gladly there are people here that want to help even newbies, like me....

believe me, i tried all info sites in the first post but none (in my humble opinion), has details or explains *everything* regarding mpc, and in my mind mpc is a bit vague...

thanks a lot for your quick answers

keep up the good work!

hold on tight...here comes another newbie question

how can i recursively traverse a directory structure doing ReplayGaining on all the MPCs, NOT HOWEVER considering each directory to be an album?

thank you!

You can use Case's Sweep.

yes but this way sweep assumes that the directory is a whole album and replaygains it in "album mode" . I don't want that.

User says in the very first post that:
"How To...
Recursively go through a directory structure and apply ReplayGain to all MPC files,
considering each directory as an individual album?

Download Sweep and Replaygain. Extract them to your path (e.g. C:\Windows).
In a command-line window (so-called DOS-box), type: sweep replaygain --smart --auto *.mpc

Already replaygained albums will be skipped, due to the --smart option. Sweep starts to scan in the current directory of the DOS-box, going through all subdirectories. To process only a specific path, use the well-known DOS commands (like d: or c: or cd..) to go to that directory (for instance, cd "c:\mpc\my downloads\") and then use the command sweep replaygain --smart --auto *.mpc to process all subdirectories from there."

plus i cant find a GUI for sweep....

ReplayGain always adds Track gain. You choose whether you want to use track or album gain in the decoder.

i don't get it....

can you please explain further?

thank you in advance....

With an mp3 file you modify the files itself to the wanted volume (losslessly).

With mpc files you just caculate how much the volume should be changed (both album gain and track gain is calculated seperately) and add the values to a tag in the file.
Then the decoder changed the volume upon playback: in the player you select if you want to use the album gain value or the track gain value.

With the following method you can use replaygain to add both album gain and track gain but the values will be the same as it handle all files individually in all sub-folders.

1) Make a batch file with the following:
2) Go to cmd and run in the dir you want to start in.
3) all mpc files in sub-dirs will be replaygained individually.

Thank you!

when i replaygain my mpc files with the batch encoder and the --auto switch, i see in the Windows Task Manager Processes Tab that apart from replaygain.exe, the mppdec.exe is also runnig. Why is that? Replaygain IS lossless, right?

Thank you!

Sure it is. But ReplayGain has to read the whole file in order to calculate the RG values. And how does it do that? Right, it calls mppdec. Decode file -> analyze -> calculate RG values -> write RG values to file.

Why is it recommended that ClipProtect be off in the WinAmp in_mpc plugin? Wouldn't it be best to have it on, to avoid clipping during playback??

Redesign of the recommended settings: I tried to make it more obvious that --quality 5 should be your first choice, and that you shouldn't go with --quality 8 when you haven't even tried --quality 5...



Provided that you use ReplayGain (as recommended), ClipProtect is very rarely needed. If you activate it with an MPC that hasn't been ReplayGained, it might lower the volume when there's clipping, which can be annoying. The best method is to ReplayGain your MPCs (and make sure you RG each album seperately, or you'll get wrong AlbumGain values).

... also fixed all broken links.

Maybe the mp3 settings thread should be modified similarly?

Good idea, i'll get right to it.

When it's not necessary, Clipprotect will not be applied. It will only make a difference in those (maybe) rare cases.
BTW I saw replaygain actualy introduce clipping at older albums (with more dynamics). But I didn't say they were audiable though.
--
Ge Someone

I have noticed with several of my albums, that Foobar2000's ReplayGain stats shows album peak *above* 1.0

Are you sure that ReplayGain will not create audible clipping? In *every* case?

Well, if you find a file where it's audible, that part of the post might be changed...

The peak above 1.0 is a result of the change of frequency components in the lossy compression, and it's common on modern peak-limited and dynamically compressed albums with all psychoacoustic lossy codecs. The same thing can happen with something as trivial as a low-pass filter. Although you're removing energy with the filter, the peak values can either fall or rise, because the peaks have little to do with the energy in the signal.

The reported peak value is the peak that will happen if NO replaygain is applied. So actually, clipping (usually inaudible) occurs regularly with lossy codecs like MP3 and MPC.

In almost EVERY such modern album out there, the ReplayGain suggested to hit 89 dB volume will be negative. This means that the peak value after ReplayGain is applied will be much lower, and almost always under 1.0

For example, if your album peak is 1.143200, an album gain of
-20 x log(1.143200) = -1.16 dB is enough to bring the peak down to 1.000000 so it's no longer clipping (assuming you have the RG preamp set to 0.0 dB (89.0 dB SPL target volume).

I have a Rachmaninov CD in .MP3 (Lame APS) that's already about 89.0 dB but clips during a full-orchestra peak. That's fine at 87.5 dB target volume (in mp3Gain) and I couldn't ABX a difference where it clipped at 89.0 dB anyway, so I could have left it as it was. I'll probably get around to reripping to Musepack some day.

... also fixed all broken links.
EncSpot link seems to be broken...

Yeah, the original site is down... using another link now.

Make sure it is obvious that the --xlevel command should always be used in tandem with the --quality setting. It is only stated in the oak-red sidenote, which some people are likely to pass over.

(this is from the first post)
Why do you use --quality 6 here? Many people will cut&paste this line right into EAC. If you want to reccommend that people use q5, perhaps you should change this example line to use --quality 5 instead. There are enough notes in this thread (both in the first post and also the replies) telling people that they should only above q5 if they notice a problem with a particular sample... but if they see --quality 6 in the example they may think "well, he is using q6, so I'll just do that too".

That's right, getting MPC, I went here to setup EAC, and wondered if quality 6 was needed. I had to remember all threads were quality was discussed, in order to decide and go down to quality 5.

So, why not? I often use --quality 6 myself. The thread is merely a recommendation; you should make up your own mind about what to use, based on the advice given there and elsewhere on HA. I wrote "example", and that's the way it's meant... use it as a starting point, do some digging on HA if you like, compare bitrates, maybe even try some ABX testing yourself. There can be no ultimate solution for everyone. And frankly, i don't see anything wrong in using --quality 6.

Higher quality seems to transcode better. Maybe this is the answer, who knows.

I rather use quality 7 for transcoding, otherwise quality 5 is fine for me.

ok, if some people write here their personal settings, I tell mine, too:

MPC --quality 7.5 --ms 15 --xlevel

according to *

Reason:
for Archival (& Listening, of course !) purpose.


q7 or q8 are still small in size (you have to compare to lossless),
but these qualities offer a quality for actions, which are normally forbidden for lossy codecs:
- Transcoding to high quality mp3 eg., or like somebody else tested: to atrac for portable MD !
- applying DSP to lossy music, ie. Logic7 / DPL2 decoder (Logic 7 for me).


MOD: no links to sharing group guides please.

I prefer "mppenc --xlevel" myself, I find standard no less transparent than any other preset. I could for example ABX standard from original at some weird case, and insane from original etc. but then again not standard from insane.
standard works fine, and from my findings it's just as good to transcode from as any other preset.

This is exactly what I did, damn. I didn't read any other posts first cause this was the 'recomended settings' post. If --quality 5 is enough in almost all cases, it should be changed.

For archival purposes a lossy codec (and even if it's the probably best one like MPC) never never is good enough. Because there's always missing something in the music-datas after having done the encoding-process. Lossy codecs are always a compromise in size/quality. This we have to put always in our minds. If you want perfection (and for archival you need it) there's only one: LOSSLESS.
My advice: Go to the extremes ! For transferring musics thru the internet, you need a good lossy codec, which generates short files: So, MPC --q5 --xlevel should be your favourite choice. And for archiving your musics for home-use take FLAC or Monkey's Audio (only go to "High"-compression, because all levels above need too much processor-time to en- and decode, without offering much more smaller filesizes).

I have added to the How to about replaygain, the part 2 about creating replaygain.txt files in a batch.:



About Replaygain a basic:

How To...

1.

Recursively go through a directory structure and apply ReplayGain to all MPC files,
considering each directory as an individual album ?

Download Sweep and Replaygain. Extract them to your path (e.g. C:\Windows).
In a command-line window (so-called DOS-box), type: sweep replaygain --smart --auto *.mpc

Already replaygained albums will be skipped, due to the --smart option. Sweep starts to scan in the current directory of the DOS-box, going through all subdirectories. To process only a specific path, use the well-known DOS commands (like d: or c: or cd..) to go to that directory (for instance, cd "c:\mpc\my downloads\") and then use the command sweep replaygain --smart --auto *.mpc to process all subdirectories from there.



2.


How to

Create replaygain.txt files for a lot albums in 1 step,
considering each directory as an individual album ?


Make a text file by an editor, *.txt,
copy following commandline into that file:

replaygain --listallreport *.mpc > "replaygain --auto album.txt"

Save this file with following name:
replaygain.bat.txt
Rename it to:
replaygain.bat

Copy this *.bat file to your "Windows-Path" (where sweep.exe, replaygain.exe etc., are located, too)

This *.bat file needs to be created only once in your life...
So don't worry...

Now the procedure:
- Open DOS-Box, go to the main folder/directory, where your albums are located, you want to add the replaygain.txt file to.
- type: sweep replaygain.bat [enter/return]
and now a file ' replaygain --auto album.txt ' is created in each album-folder.
That's it !

Added --xlevel to all recommendations, simplified some other stuff.

I Splitted the tag question to http://www.hydrogenaudio.org/forums/index....showtopic=11596

Decode question splitted : http://www.hydrogenaudio.org/forums/index....showtopic=11645

TO ADD:

LINUX RESOURCES:

players: XMMS, BMP, LAMIP
xmms-musepack plugin
lamip-inputmpc plugin
bmp-musepack plugin

Debian Packages:
http://rarewares.hydrogenaudio.org/
http://www.hydrogenaudio.org/forums/index....showtopic=10340


GRIP FOR *NIX (v3.x.x)

(will be put into an actual guide, soon...)

(for scsi emulation of ide)
CONFIG>CD
cdrom device: <scsi device> (i.e. /dev/scd0)

CONFIG>RIPPER
Ripper: grip (cdparanoia)
(disable) Disable paranoia
(disable) Disable extra paranoia
Disable scratch:
(disable) detection
(disable) repair
Rip file format: ~/%A - %d - %t - %n.wav (e.g. "led zeppelin - coda - 03 - i can't quit you baby.wav")

CONFIG>ENCODE>ENCODER
Encoder: other
Encoder executable: mppenc (or mppenc-static, ~/bin/mppenc-static, /home/user/bin/mppenc-static - make sure executable is in $PATH or an absolute path is given)
Encoder commandline: --quality 5 --xlevel --artist "%*A" --title "%*n" --album "%*d" --year "%*y" --track "%*t" --genre "%*G" %w
Encode file format: ~/%A - %d - %t - %n.mpc (e.g. "led zeppelin - coda - 03 - i can't quit you baby.mpc")

** CURRENTLY, XMMS MPC PLUGIN DISPLAYS TITLE, ARTIST, ALBUM CORRECTLY
** full list of switches here: http://www.nostatic.org/grip/doc/gripswitches.html

CONFIG>ID3
(disable) Add ID3 tags to encoded files (preference, id3v1 is possible)
(disable) Add ID3v2 tags to encoded files (Musepack does NOT use id3v2 tags!!)

[edit: removed image URLs as they don't exist any more ]

later

how do i on the fly encode with cdex?

thx

CD Drive -> Ripping method: Paranoia, full

Generic -> ID3 Tag Version: None

Encoder -> Parameter string: --quality 5 --xlevel --artist "%a" --title "%t" --album "%b" --track "%tn" --genre "%g" --year "%y" %1 %2

Encoder -> File extension: mpc

Encoder -> On-the-fly encoding: [x]

CDex configuration

(Courtesy of Volcano)



P.S. "My Favorite Lame Preset alt-preset fast insane":

There is no "fast" insane, only --alt-preset insane.

that only works with pre-1.05 versions of mppenc.

You mean the on-the-fly part? Sorry then, I don't use CDex myself...

The link to the Winamp plugin is dead.