How do I go about removing and undoing the changes created by MP3Gain in my mp3s. The only evidence I have of the changes is the .log file created in the directory of MP3Gain which shows the changes to each individual file.

Short of doing each track manually, how can I undo this mess I've done (apparently, even at 89.0db, each track had clipping still, but I still let it continue).

Thanks for your help.

voltron

you can't. obviously the program should have some sort of history feature that rolls back changes, but it doesn't

You can album-gain the albums up to where you thought their average gain was before, but if you used track-gain, you've messed everything up pretty bad I'm afraid. The only option is to manually change them back in that case.

When you have the log you can reverse it manually. You can use the constant gain to reverse the gain you applied ealier.
If fx. your log says you changed the gain by "-2" you set constant gain to "+2".

edit: but there's no easy way to remove the gain you applied. It's not written to tags like for ogg vorbis and mpc.

It's planned that MP3Gain writes the dB changes into custom ID3 tags. They can be used to automatically reverse the changes later.

It's also planned to use ape2-tags in mp3gain.

In summary, doing the constant gain opposite to what my log file says will restore the gain back to what it was before MP3Gain, correct? I just want whatever I had before back, as it was, without any changes whatsoever. Thanks Jan S.

voltron

PS
I'm having some problems with the following files:
R:\mp3\live\comedy\abbott & costello - whos on first.mp3 -16.6
R:\mp3\live\rock\vines, the\the vines - ms jackson (live).mp3 -18.1

First of all, the slider only goes up to 12.0, and (B) it's only in increments of 1.5.

This would be a great compromise for portable mp3-players, because they probably will never handle ape-tags and a tag-based gain-info (like ogg vorbis, mpc).

Haha, defeated, completed in less that 10 minutes.

For future dorks that f*** up their files without knowing what they were doing:
(1) take log file and copy paste in excel
(2) sort the gain column in numerical order
(3) copy each set of gain'd mp3s into an m3u
(3.2) load m3u into mp3gain!
(4) constant gain that m3u set with a gain of -1 * (gain listed in second column)
(5) do with every m3u until completed

Thank for your help guys. =) Appreciated Mucho!

Glad you figured out the Excel thing. That's what I try to suggest to people in this situation.

What puzzles me is that your tracks were clipping at 89dB. It's true that _some_ tracks will clip at that level, but I've found that the majority of them don't.
What kind of music were you doing?

-Glen

It is not exactly on topic, but I do not want to open a new thread for that .
I got a file on which Mp3Gain is not able to determine the peak level. I only says '????'...what is this supposed to mean?

Allow me to fix my sentence, maybe 100 or so trakcs had still been clipped it had a Y in the Clipping column, so that's what I went by. The mp3s are all created using different encoders over the years (mostly LAME) and mostly rap/hip-hop music.

It means the mp3 has some corrupt data, and the mpglib code is returning a peak of over 1,000,000 (where max should be 32,767).
I have plans to put Peter's improved mpglib code from foobar2000 into MP3Gain, so that it will handle corrupt data a bit better. That particular plan is currently on my "To Do" list after the much-needed "store info in tags so you can undo, and so you can analyze only once" code

-Glen

glen,

awesome idea about undo
if you do get peter's edited mpglib, can you somehow make a dll out of it so we can us the same decoder on other programs?

i'm making an mp3 player that is just a notepad like foobar but no buttons no display options, just directsound and play/pause/next, wouldn't midn using the nice mpglib for my little monster

That is interesting...the file seemed fine to me; no errors or clipping at all.

If you can get hold of the original mastering of The Wall, try that. Very many of the tracks will clip at 89dB, I remember one of them even gets pushed to >120000 peak value when amplified to 89dB (you don't want to know how bad it sounds without a hard limiter ).

(Of course, one wouldn't want to apply Track Gain to the album because many tracks fade into each other, but even with Album Gain (around +2 dB) some tracks clip hard. Just an example, anyway... this will probably happen on any dynamic classical CD too.)

Ehum,
I don't like the 89 dB settings either.
To get best possible sound on my mp3 hardware player i use the "Apply Max No-clip Gain for each file"
option. It's my favorite choice. This is a track gain option and will not keep any relative volumes between tracks on an album which is the main intention of the author of mp3gain.

But that's how I want it.

The result of test I have made is that many low volume level songs gets more volume and the ones having too high volume gets lower volume in a nice way. You might have to alter the volume on the player to get the same average volume level on different songs but i haven't been needed to do that on my MP3 CDs.

Don't do this on very large collections as on your entire hard disk. If the music genres differ much like having classical music in the collection or continuous music pieces without gap between them for instance it's a BAD idea having the volume altered "track gain" wise. Low volume is often "wanted" on some tracks on classical music pieces.

I can add that all the music I listen to is of the same "narrow" music genre and that's when "Apply Max No-clip Gain for each file" is a good option. If all your music is Hip Hop it might be OK. If you do this you might like the result but you will probably NOT be able to get your original levels back after having changed the volume again.

Always copy some pieces to a new directory and test first to see if you like the result. Burning the library before altering can be a good choice.

Unfortunately if the gain is altered (mp3gain used) it will be easier to re rip and encode again to get the original volume back if the collection is small.

I personally would not recommend using mp3gain if you will play your music on the computer. Use a player or plugin that adjusts the volume when playing instead then. The sound on my computer is awfully bad compared to when listening from my CD player so the eventual increase in quality will probably not be worth using volume altering (mp3gain) either unless you have a very good (=expensive) sound card and HiFi equipment.

Conclusion don't alter the files if you will have them on the computer. It's hard (impossible) to have the original volume level back after having used mp3gain, so consider twice (seems too late for you this time though) before using it if it is your goal to alter the volume levels.

/ Alexander

I thought that mp3gain applied the gain to the mp3 itself and not to a tag...
Am I wrong?

No you are right. Did I say that, sorry if I did.

I'm no expert on how it's possible to have the players gain adjust the levels automatically. But you are probably right the gain info has to be added in a tag first if you want to use that.

/ Alexander

That's OK if you'd like every file to be as loud as possible, but it's not what ReplayGain is intended for - ReplayGain is designed to make files equally loud. Sure, if the songs on your player are relatively new and are of the same genre, the volume will still be roughly equal (because everything is made as loud as possible), but as soon as you mix old with new music you will get large differences in volume which defeats the purpose of ReplayGain.




This is not true. Even the cheapest setup (like I have here ) should be able to handle 89dB files without a problem. You've just got to turn up the volume a little more - not so much that you'll experience a significant increase of noise though.




You mean, a compressor? :x That totally destroys the sound because it takes all dynamics out of the music.

It's all right for parties, but...

Just my 0.02€

That was just what I was trying to say. But now it's probably clear enough for everybody.
I think you must have missunderstood what I meant.
Tried to explain it but maybe the result is too much text so the point is lost. B)

The point was that voltron has clipping as it is now.
My solution would at least fix that without having to do a lot of work. I would not suggest it otherwise. I know it's "wrong".



Well yes but if you use mp3Gain your original volume levels will be more or less impossible to retrieve. Is it worth it. If you use a "replaygain plugin" solution you still get good listening volume level like you would have got using mp3Gain and also you don't lose the original volume levels and that's what voltron (this threads first question) wanted to avoid.



No I don't mean compressor plugins. Must have missunderstood me.
I know I was quite unclear about this sorry.

/ Alexander

I don't understand why people are so keen on preserving the original volume levels. It can definitely not be called "artist's intention", simply because nobody can anticipate - let alone dictate - with which volume the CD is played.

So my question is, if a) most of the newer CDs are mastered in a wrong fashion (-> clipping), and b) YOU are the one to decide the playback volume, what would you need the original volume for? Mind you, we're talking about lossy audio.

Well if one song fades into another. If you don't preserve the volume levels you will get an instant volume change here. Is this what you want then. And if the music has clipping from the beginning there is nothing we can do about it but if we alter the volume again how is the quality going to be better?

Also if you want to do a comparison test between original CD and the encoded mp3 the relative volume levels (as well as all sound on the CD) should be an exact copy of the original as possible. For non picky people this is of course not of so big importance and as I have said earlier I'm not following this filosophy either. Overtwise I my suggestion of using "Apply Max No-clip Gain for each file" would be totally wrong. But at least this choice is not gonna giva a worser quality (when listening to the song).

To keep the relative volume between the songs is despite of this what the inventor of mp3gain has in mind and that's the program that voltron has used. Also he wanted to have the original volume back (yeah i say this once again). And I don't know why he wanted to do this?

So I agree with you more than I disagree. Still I understand the other point of view here.

/ Alexander

Alexander:



Ah, I see. Sorry for the confusion.

But: He said he was processing rap music, which, typically, is mastered very loud and thus needs to be reduced by MP3Gain by quite a large value in order to get a 89dB volume level. If a loud (> 96dB) track still clips at 89dB (similar to what Garf reported happening with a Red Hot Chili Peppers album), then it would become even quieter if you applied the maximum no-clip gain, so it would be even further away from its original volume.

I hope I haven't misunderstood you again, because I've got a funny feeling writing this.




OK. Again, sorry for the misunderstanding.


Regarding your reply to CiTay's post:



That's what Album Gain is for.




It will be "better" in the sense that at least the additional clipping caused by the MP3 decoding process will (in most cases) be removed.

But, again, ReplayGain's aim isn't improved sound quality (because, elimination of lossy audio clipping aside, it's not possible that easily ), but making songs or albums sound equally loud.

CU

Dominic

Why don't people just use the "Apply Max No-Clip Gain for Album" option? This would make it so that there is no possiblity of clipping, and so the whole album has the same relitive volume for each track as the origional. What is the point of using an 89db album gain? And is there any way to find the max no-clip gain on an album basis using the VorbisGain / MPC Replay Gain utils? Thanks.

Edit - For Vorbis / MPC, I have been applying a constant -3db gain for encoded albums (Vorbis: -3db in Winamp plugin || MPC: replaygain -3). From what I can see in Nero WaveEdit, this removes 95% of the clipping from my files and keeps them at a decent volume. I still wish I could use a max no-clip gain for each album... any help would be greatly apprecieated (MPC is the most important to me - it is what I use to encode all my new albums.) Any solution must be able to work with Winamp 2.81 w/ Vorbis v1.2.9 & MPC v0.98 plugins.

I think that the main reason is a human ear which hear different frequences with different loudness. So, if your album includes songs of various music styles or just mix of new and old songs, the difference of loudness between songs will be significant, even if you use the same level of recording volume. This is why, if I understand correctly, MP3Gain was developed for. However, I've noticed another thing. It seems to me that one of my records (it was a song by Cream from early 60's), processed by MP3Gain, sounds a litle bit different compared to the original.

The question to developer: Could it be that MP3Gain has changed the frequency range of original song by changing the volume of different frequences within the song.

Maybe i may answer: No, that's impossible. MP3Gain modifies the global gain field of each MP3 frame; the overall loudness. No other values are touched. Read the MP3Gain help for further details.

Just a reply to VIPER1313s comment.


1. If you listen to your files on a computer that can use Foobar the best is to use Foobar and not alter any volume according to me.
It has both Apply Max No-Clip Gain for Album AND for Each File as options. Use the one you like and you can change your mind afterwards because the original volume is not altered.

2. In my case I listen to the files on a mp3 hardware CD player. Then also the above 2 choices are the best but you must use mp3gain. Which you choose of "for Album or for Each File" is up to your taste.
I use "for Each File" because i noticed that the files on a "The best of" album had VERY different volume.
This might be because different mastering techniques / philosofies where used at the different original recording "years".
Anyway I'd rather like to have them the same volume and my choice fixes that.

The downside of this choice is that overlapping album songs may get different volume.

But the overlapping songs are usually mastered whith the same technique so the difference is usually like 1.5 dB after the use of MP3Gain "Apply Max No-Clip Gain for each File" which is hardly noticeable. As a bonus the song that otherwise would have got 1.5 dB less volume now will play a little louder which makes the signal being further away from the noise floor of the player.

OK that's why I used the "Apply Max No-Clip Gain for Each File".

/ Alexander

well, to "remove" the quieting of the mp3s, you could simply Apply Max No-clip Gain for each file choosing the correct option of Album of Track gain, depending on what you did before. Because this is not clipping, you are NOT losing anything, AND, by maximizing, you are probably bringing the individual file to file loudness relationship back more to where it was before....at least more so than the equal volumes...I understand how this would not be true for very quiet songs originally, because they would now be MUCH louder before clipping...but it would probably help some...ah, but the mroe that I think of it, the more I tell myself, that all this will do it bring all music to a pointless loudness, because I am sure the gaps between clipping and not widely range from recording to recording...hmm, maybe someone can sort out what i mean here, and attach some hard facts to help...