I have an Iriver H140 [love it]

Ive just ripped all my cds using Eac and lame

I want to alter them so they are all roughly at the same volume level

The reccommended "target normal volume" setting on mp3gain of 89 db is way too low

Ive scanned some of my ripped albums and they tend to range from about 89 ish up to 98 ish

I want to make them all around 95 db

When I do this a lot of them show up as clipping

Surely this isnt going to spoil the sound that much

Also is there any way to get over the different requirements for listening to albums at the same relitive volume setting
And listening to play lists from multiple albums

thx

I suggest you read the information provided on www.replaygain.org. The ideal is to reduce or eliminate clipping. Turn up your volume nob, then you'll get the "loudness" you want.

Also in your future posts will you please write normal sentences that end with periods, etc., and are not spaced apart after each line.

no wonder they clip at 95, that's still pretty loud, and the tracks are not too unlikely to clip. however, the clipping detection of MP3Gain is very very sensitive, you may not hear the clipped samples. it depends how good your hearing is and how trained for this kind of disturbance.
anyway, you might consider setting the gain to ~92.

please rephrase.

edit: spelling

After ripping loads of the cds were higher than 95 db in any case.
So if anything Ive improved at least some of the mp3s.

The iRiver H140 is quite a low volume machine.Even at 95 the volume at the max setting is not very loud.
To rephrase the last part of my post:
You can adjust gain by album or track.

Album setting is good for listening to whole albums.
But not so good for listening to playlists made up of different albums.
True or false.

I will try and improve on my typing skills.
Btw nob is spelled Knob.

Please note this article:
http://pages.globetrotter.net/bakerywizard/guide.htm

He implies that at 98 db the clipping is not noticeable.
He could be a little bit deaf or untrained in hearing defects in mp3s.

A bit like me.

No offense intended.

it depends on what you want. read the FAQ inside MP3Gain.

That guy needs to be flamed for spreading dangerous misinformation like "I don't hear any problems, so I recommend fucking up your music collection with the same insane replaygain reference level." Chances are that he doesn't really understand the concept of replaygain or the motivation behind the "low" reference level.


He could have a collection of contemporary pop and rock music which is mastered so loud that a 98dB reference level will STILL always result in a level reduction for playback so no clipping is introduced. However, throw one record from 1990 at it, and chances are that it'll go BOOM. (or rather CLIP)

The problem is this.
If I set all my albums to 89 db gain they sound too low even at maximum volume on mYy iRiver H140

Hmm
I might try 192 db and see how it works out

Just done a quick check

This is what EAC and lame provided for these albums

Astral Weeks Van Morrison

Approx average vol = 90

Whats the story morning glory


Approx average vol = 96

Hmm

What headphones are you using? Perhaps your iRiver can't drive the properly, in which case you should invest in either more sensitive phones or a headphone amp.

Sennheiser px200

Lots of ppl complain about the low volume issue on the iRiver H1xx series

choccyx,

Set the target level at whatever you want - if it sounds fine to you, it is fine. If all your songs were loud to start with, then it won't be a problem. If you're maxing out your player's amp and headphones, then a little extra clipping may be the least of your problems anyway!


However, as for http://pages.globetrotter.net/bakerywizard/guide.htm



Well, the best you can say about this is that at least people can easily undo this advice if/when they discover that it's wrong.

I love the assertion that no one has done a listening test - would you like to listen to this? ...

http://www.hydrogenaudio.org/forums/index....showtopic=29512

Shall we ABX? ;-) You can make your own original by mp3ganing back down to a sensible level. The recording is from 1937, and pushing it to 98dB predictably wrecks it.

Maybe I'll email the author of the site - you never know.

EDIT: Ah, but I feel I should try his software and say something nice about it first. It looks very cool - if only it interfaced with foobar!

EDIT2:


On second thoughts, I think I'll leave it to someone else! Be gentle.

Cheers,
David.

I use the px200 on my imp 550, and music already sounds loud at 21/40 even if I'm sitting at the back of a bus, where the engine usually is. 40/40 would be insane, unless the low volume issue is really serious.

Edit: The mp3gain level for my mp3s was 94 dB.

I have mine set to 95 gain at this moment in time.
I must say it sounds sweet and I cant really hear any clipping

Im gonna use this for a while and if its ok Ill stick with it.

89 is way too low for me.

And what about album gain max. no-clip to each of your albums...
I think that your goal is impossible when 89db is too low for you AFAIK...

And what about album gain max. no-clip to each of your albums...
I think that your goal is impossible when 89db is too low for you AFAIK...
[/quote]

When I set up "Enable Maximising Features I get informed that this is not a good thing to do


http://replaygain.hydrogenaudio.org/faq_norm.html

Thx

I got an iPod mini and all my mp3's are gained at 89dB...I use Sennheiser Px-100 headphones and I never had any problems. Even at the airplane there is still enough volume avilable...never had to turn it all the way up to the max.

Well what can I say? Just try it! In most cases when you apply max. no-clip album gain you lowers the volume of the original MP3 (reffered to 1980-today rock / metal albums)...

I (a newbie) have a question about all this:
If I rip one of these modern, loud, and extremely compressed cd tracks, encodes with LAME and get clipping. Will not these "clipped" parts in the track sound bad, whatever you do to adjust it after encoding?
I mean: can MP3gain or anything else repair the damage?
Edit: In this example there's no clipping before encoding, it's introduced during encoding. I've done som testing.

The only real danger lies in applying Max No-clip Gain for Each File, which is rarely a wise thing to do.

But if the relative gain of all the tracks being processed as an album is suitable to begin with, you can never go wrong with applying Max No-clip Gain for Album.

Regards,
Madrigal

Yes, by Mp3Gain you can repair the clipping that was introduced by encoding...

Yes, it seems that Mp3Gain is doing a great job.
But don't I lose information when clipping is introduced by encoding?
If I lose data, can Mp3Gain really undo it?
I've read the Mp3gain FAQ, can't find the answer there. In fact I hope I'm wrong, and no info is lost. But I want to be sure.

I was surprised that, in five minutes, I couldn't find the answer to this on-line either. It must have been answered many times, but it may be quicker to answer again than to keep searching!

If something is "clipped" it means that the sample value is (or should be) above digital full scale - but gets "clipped" to exactly digital full scale, because there are no larger numbers available to represent the true value.

The mp3 encoded version is clipping because the original CD had sample values were very near, or equal to, digital full scale. Mp3 encoding changes the waveform slightly - so that (often as not) these nearly clipped samples get pushed slightly higher, into clipping.

The reason you can recover these samples is because mp3 files can store values greater than digital full scale. So, they've been pushed up out of the range the mp3 decoder or player can handle, but they're still in the mp3 file. If you can adjust the volume down slightly before you play the file back, everything will come out fine, because the values will be decoded into the range below digital full scale = no clipping.


If mp3 encoding only pushed the values slightly higher, it probably wouldn't be worth worrying about this, but some circumstances cause peak values to increase dramatically - it's not a fault, just a fact of life - there's more about it here:

http://www.ff123.net/norm.html


To prove to yourself that what I'm suggesting is happening is happening, try this:

1. Take a file (preferably a nice quiet file) and mp3gain it to some stupid amount (110dB or something).
2. Listen to it - it should sound awful - it'll be full of clipping.
3. mp3gain it back down to, say, 89dB.
4. Listen to it again - the clipping has gone. It was only temporary.

Obviously mp3gain can't do anything about clipping which was present in the original source (e.g. CD).

Cheers,
David.

I have nothing to say you more... I've already answered. Sorry for my poor English but I think I've written it very clearly...

2Bdecided, thanks very much for your full explanation!

I didn't know about this, thanks again for explaining!
The last month I've read a lot of posts here at HA, but haven't seen this explained before. (Or I missed it )

Your english is good enough for me. I was only hard to convince, I'm sorry for that.

Back to the bakerywizard site linked above.

There's been some private emails, where the bakerywizard author demands ABX evidence, using appropriate music.

I took the sample Enola Gay (I think it's by OMD) from a codec test (a 64kbps one IIRC). I encoded it using FhG 256kbps FastEnc (of all things!), made a copy, mp3 gained the copy to 98dB (which required a 7.5dB level increase).

I decoded the untouched and gained mp3s, reduced the level of the latter by 7.5dB, and loaded them into foobar ABX comparator.



It was actually reasonably hard - the track is already slightly distorted, but gaining to 98dB makes it more so, and changes the sound too.

I can quite see how, if you were a DJ, you couldn't care less.

However, it is audible, and it'll wreck certain tracks.

(I didn't go looking for such a track - Enola Gay was the first "pop" track that came to mind that fitted the ABX request - old but not too old, and on a test CD I had with me)

Cheers,
David.

To corroborate 2Bdecided's findings (I hope my methodology is OK):

I encoded the same enola_gay sample with Lame 3.90.3 --alt-preset standard.

Replaygained a copy to 98db using mp3gain.

Decoded both mp3s with Foobar2000 (no DSPs are enabled).

Loaded the replaygained version into Soundforge 5.0b and reduced the volume by 7.5db using the Process > Volume menu (this is so we can perform a fair ABX test).

I chose to do 8 trials and found the first 2 or 3 trials quite hard. After that I could pretty much pick out the more distorted version quite easily. The screenshot shows the timing of the area I concentrated on.



Suggesting to use an mp3gain value of 98db is bad and irresponsible!

Ok... I'm the guy how posted that misinformation regarding replaygain setting.
I can see now why this is definitely not recommended!

The only reasons I can think of that steered me in the wrong direction right from the start are:
To me it didn’t make sense to take a perfectly good track having an initial 98db level and bring this level down to a faint 89db.
I was under the assumption that since nearly all new CD's are produced at similar high levels nowadays that old CD's (or their resulting mp3's) should be brought to closely match this level instead of the other way around ??
Well all clipping issues aside, that would make more sense I guess but I can see why it should be the other way around now.

Also... after having previously used MP3Trim to adjust some of my albums in the past, I had noticed that its "0" db target reference corresponded almost exactly to a 98db target value in MP3Gain! Why such a huge discrepancies between the two application’s target values? I specifically remember writing to both authors at the time to find out but never received a reply from either ones.

Well thanks for taking the time to steer me in the right direction anyway. I'll be correcting the information I have posted on my site soon.
Cheers!

Andre Aylestock
Author of PartyTime Jukebox

Result! btw, (as I think I said before I knew you were reading) your software looks very cool.



That's an easy one. MP3Trim reports the peak sample value. Most pop tracks will peak at 0dB FS. If you load a loud track which is already about 98dB SPL loud (as ReplayGain measures it) and peaks at 0dB FS (as MP3Trim measures it) into both, it'll seem equivalent. If you load a track that peaks at 0dB FS but is about 83dB loud into both, they will seem equivalent!

The discrepancy is because they're measuring two very different things.

Cheers,
David.

Thanks for taking the time to listen and getting back to us here.

Thanks, I am also trying to make it fb2k compatible in the future but this might be tougher than anticipated.
I have a few commands that are working fine with winamp 2.x but have not yet managed to duplicate in foobar. Specifically the ones dealing with the playlist.

I've posted in fb2k development forum looking for some feedback so I'll be checking that next!

Again, thanks for the info you've provided!


Andre

I am always happy to see webmasters edit their page when someone points out a mistake.

Hey I'm interested in using your PTJB - when you've mastered Foobar. (I do not use winamp).

What useful and important information this is!
I don't know why this is not in mp3gain FAQ.
Now I understand what mp3gaining is for.

Madrigal: "The only real danger lies in applying Max No-clip Gain for Each File, which is rarely a wise thing to do."
Why is that dangerous? it sounds like the best thing to do, cos i want each file to play as loud as possible without clipping.

Could you please explain this to me too. I tough it was the best of both world, making my mp3's the loudest possbile witouht having clipping ??

Soren

2 Keithboy & Soren: The problem lies in the fact, that the track max. no-clip destroys the balance among the songs of the album, so eg. a quiet track (like some intro, intermezzo, outro) is abnormally loud and it ruins the listening pleasure to the whole album...
I hope you know what I mean!

oh ok i thought you guys meant that there was some kind of other problems occuring when using the the "max no clip" option instead of "normal" gain, be that track-gain or Album-gain. (oops that was a long sentence!) other than that i understand what you mean.

The only problem is that you'd have to change the volume all the time since some songs don't allow anything louder than 89dB, others are fine with 98dB. Also, sometimes you have continuing tracks that are split into 2 tracks...and therefore you might experience a drastic volume change...

Thanks for the explanation k.eight.a , and i must admit that it is something i never tought before !

Soren