Well I started using EAC and used MP3gain to increase the volume on my older CD's. I don't know if it's just me or what, but it really seems like the MP3gain helps the old CD's sound alot better. They seem livelier with a little more kick. Is it just me?

It's just you.

Seriously though, what you've done isn't really any different than just turning the volume up on your stereo.

MP3gain alters only the amplitude of the signal. Any perceived difference other than in volume is placebo.

Not really. An increase in volume is known to sound better, provided there isn't any increase in distortion/noise/painful loudness/etc.

MP3Gain actually changes the MP3 rather than just adding a tag I think.

Replay Gain, Vorbis Gain etc. are great, but not to make old stuff louder... it's to make new heavily compressed shit quieter. This way you can keep your volume control higher which makes the older music sound great but the newer crap won't kill your ears.

Both ReplayGain and Mp3Gain have a very similar goal: Bring all Tracks/Albums to the same perceived loudness without losing quality. If a track/album is more quiet than the target loudness, the volume is raised, if it's louder the volume will be lowered. The difference between Mp3Gain and ReplayGain ist just the method used to raise/lower the volume.

ReplayGain relies on a player reading the tags and lowering the volume on playback, while Mp3Gain (losslessly) changes the mp3 data itself to adjust the volume. The result when played back should be (approximately) the same with both methods.

But isn't changing the MP3 data destructive? Or can MP3Gain losslessly restore files back to their original state? I know fooobar2000 says that if it applies gain directly to the MP3 data is irreversible.

Like mp3gain its lossless , but cannot be undone.

The difference between MP3Gain and foobar2000, is that MP3Gain stores undo information in APEv2 tag. This way you can revert losslessy to your original file. foobar2000 does not store undo data, but changing the amplitude of mp3 blocks is limited to 1,5db steps, so theoretically you can revert all files to their original state, by trial and error. There's another thread inhere that explains how you can check the CRC checksum of the mp3 data if the file was encoded with LAME by trial and error, but really... I don't think it's worth it


It sounds like you choose to gain your files above the recommended 89db. You should be aware that this might result in clipping when these files are played back. Your files should sound better once you turn the volume knob instead

You could manually change the RG tags and reapply gain to get the old volume back.

[quote name='hardrockin80s' post='516621' date='Sep 15 2007, 00:47']
They seem livelier with a little more kick. Is it just me?
[/quote]
It sounds like you choose to gain your files above the recommended 89db. You should be aware that this might result in clipping when these files are played back. Your files should sound better once you turn the volume knob instead
[/quote]



I burnt a copy of song with the mp3gain and one without. I cranked the volume up on the one without and left it lower on the one with. The one with mp3gain had a more of a solid sound to it. and yeah i will admit i i adjust it where there should be some clipping, but its nothing i ever notice.

You should have done the comparison double-blind. "Solid" is meaningless if it can't be described in terms of time/frequency/loudness.

By double-blind, I mean a situation where both the listener (you) and the person who chooses the CDs (someone else, preferrably) have no idea beforehand which CD is going to be played.

You might need some helpers for this, but believe you me, with a thorough double-blind test, the kind of differences you are describing will not be noticed.

Or maybe you should have made the MP3Gained ones more quiet and the non-MP3Gained ones more loud. Louder does sound better, after all.

Maybe we should explain exactly how mp3gain works so that he may realize that there is no magic involved.

Yeah congratulations you've just increased the volume, it kicks butt doesn't it. Here's a tip though, it works exactly the same using the player volume control.






Yeah, if you ever doubted the existence of the placebo effect then all you'd have to do look at some of these posts where people are listening to exactly the same thing (without realizing it) and start marking these very definite claims about improvements or degradation.

How many posts have you seen where someone switches from one bit-perfect software mp3 player to another and then starts raving about the improvement. Or tells you how much better there music sounded when they switched from FLAC to WMA lossless, or how much worse FLAC sounds compared to the original wave file and on and on it goes. No wonder it's hard to take people seriously without ABX tests to back up their claims.

Volume changes (or the nice effects of volume changes) are ABX-able. Sometimes even differences smaller than 0.5 dB differences can be noticed, if I believe some of the previous talk here at Hydrogenaudio.

It's quite possible there was a real audible difference in audio quality, not just placebo.

The crucial comparison would be, like I said, a comparison with MP3Gained songs more quiet and the non-MP3Gained ones more loud. Louder does sound better, after all.

Hey, you guys leave hardrockin80s alone! If me and Glen have a chance to sell mp3gain as an audiophile audio improvement tool, we're going to take it!

Shall we bundle it with a green marker pen or something?!

Seriously, all other things being equal, louder is usually perceived as better. Down to a very fine degree (I've been tricked with 0.5dB!).

However, while mp3gain can't remove distortion and clipping from a source, it can prevent extra clipping+distortion during decoding. If it's used to reduce the loudness, it can make things sound better that way. (Though most people prefer louder+more distorted to quieter+less disorted).

Cheers,
David.

Yeah I know that volume changes are ABX-able. My point was that the OP obviously thought something else was going on here beside simple volume increase when it wasn't. Well at least so long as he wasn't increasing the volume to the point of clipping that is.


Well actually there is one condition in which upping the volume with mp3gain *can* make a real improvement above that which would be obtained by simply turning up the volume. This is on some very cheap and nasty players where there's a fairly high noise level and where the noise level is linked to the device volume setting. I've got a cheap little mp3 player like this, it really is advantageous to have material at 98dB instead of 89dB, but I don't think that this is the effect that the OP was referring to.

First of all my system isnt cheap. I have JBL speakers in my truck and an Alpine Head Unit. Second of all I thought we all came here just to discuss things and not to get punked for an opinion. All I did was state my opinion on something and from reading some of the responses on here, its like your trying to make me feel retarded or something. Granted I dont know half of what most of you do, but when my ears tell me one thing sounds better than the other then thats all that counts to me. Isn't that what alot of you say everytime someone asks you whether to use aac,ogg or mp3?

First of all, we love to discuss things, but we do have a policy here at Hydrogenaudio where opinions need to be backed up by facts (double-blind ABX tests). It's in the Terms of Service, as item number 8. You agreed to that when you joined this board, so please take this into account. Thank you.

Secondly, you might feel frustrated, but you can't expect everyone to agree with you all the time on everything, I guess. Please note that there's almost always something new to learn in anything you do, even if you are an expert.

----

I don't deny that your louder songs sounded better. In fact, I think it's perfectly normal. It's just that MP3Gain isn't the only way to make things louder...

(edit: typos corrected)

hardrockin80s, you may think you're hearing a difference but it's just your mind playing tricks on you. mp3gain does nothing to your mp3s than make adjustments to the global gain field for each frame and each frame receives the same adjustment. What exactly does the adjustment do? It changes the volume!

I think this is a pretty fair test for you to try:

• Take a lossless file, encode it to mp3 and apply gain using mp3gain.
• Take the same lossless file and apply the same amount of gain to it and then encode it to mp3.
• Try to ABX the two mp3s.

greynol: that's a bit misleading. Differences in volume are ABX-able, and hence quantifiable by people.

Louder sounds better. That might be your mind playing tricks on you, but the loudness itself is not "your mind playing tricks on you."

-----

And no, I'm not condoning the loudness war in any way!

Of course you are right, but it's pretty obvious that hardrockin80s is attributing more difference than just volume; or are you now ok with his descriptive term, "solid"?

I think it would be more useful if you'd pick apart my proposed ABX test.

Yes, I'm OK with people using the word "solid" to describe the effect of loudness, since it can be attributed to either time, frequency, or intensity.

I don't think your test is flawless, because of ATH curves.

edit: toned down scepticism

IMHO a better test would be to...

1. Get an MP3 sample with some headroom for amplification and make a copy
2. Apply +3 dB gain to copy, using MP3Gain
3. Compare "original MP3 with +3 dB preamp on playback" with "MP3Gained MP3"

Let you skepticism shine.

I didn't think it was flawless either, but I think it should be adequate provided the guy isn't applying more than a few dB worth of gain. I certainly think it would dispel the magical attributes that are being given to the mp3gain process that extend beyond simply turning up the volume.

What would you say if the test was conducted and he failed? The affects of ATH and other things like clipping will certainly not make the two mp3s sound more similar!

EDIT: I like your proposed test, but is there an ABX program that can do what you want? I don't think this is achievable with foobar2000.

Failure to ABX never disproves a difference. It only means that there's insufficient grounds to suspect that there is a difference.

You can apply preamp (through the RG dialog) in foobar2000's converter.

What I'm saying about ATH is, there should only be one controlled variable in a scientific test.

Yes, kjoonlee, I'm well aware of that. Let me know when you're done with your nitpicking.

A failed ABX test should be more than adequate to demonstrate that mp3gain isn't magic.

Yes but it doesn't render the test useless provided that people don't make the wrong conclusions. It was only intended to be an eye-opener.

So, why don't you document your proposed test so that hardrockin80s can demonstrate to himself what mp3gain is really doing?

The purpose of this forum is to share knowledge and be enlightened. I'll nitpick if I have to, when I offer constructive criticism.

I already have. If you want, you can fill in the details. Please go ahead.

I can't because I still don't think you can apply a preamp to one file and not another when performing ABX or are you suggesting a different way?

foobar2000's converter can apply preamp during conversion.

(RG settings.)

I suspect foobar2000's DSPs work the same whether they're used for playback or conversion.

Actually, I think this can be done double-blind with two instances of foobar2000 running. One with +3 dB preamp and one without, with people taking the roles. No ABX software necessary, yet ABX is possible.

I'm not really frustrated I guess I just don't understand why everything has to be ABX before it's considered legit on here. It's not rocket science it's just listening to some good tunes...lol and please believe me, I'm not trying to get everyone to agree with me on here.

I think we're all trying to get you to understand that mp3gain is essentially doing nothing more than applying volume. Through a carefully designed objective test you may get closer to the truth.

Because the human brain is very good at noticing things that aren't there. (At least it's better than getting eaten alive for not noticing things that are there....)

Consequently, we need to have a way to make sure that a perceived difference really is a difference in what your ears hear, not just a difference in what your brain thinks it heard. We also need to make certain that there's no possibility of chance or luck involved.

Thus, we have double-blind tests with checks against the p-value.

Or, to put it another way, we take audio very seriously, using the same approach that people use when they take medicine very seriously.

When new drugs are tested, procedures are put in place to remove the well known placebo effect from the results.

Likewise, when we test audio, we insist on procedures to remove the well known placebo effect from the results.

Many audio discussions don't bother with such rigorous procedure. Some branchs of "medicine" also despense with this procedure: they're the braches of "medicine" which only work if you imagine they do!

People who really want to be cured trust proven medicine. People who really want better sound trust proven audio.


I'm the author of the algorithm used within mp3gain. It does nothing, by itself, to improve sound quality, apart from the scenarios I discussed in my previous post.

Now that you've explained your listening equipment, mp3gain may well put the sound through various parts of your system at a different level, and various non-linearities in the system may (probably do) mean that it will sound different. But it's not magic. It's just a volume control, at a difference place in the chain.

Cheers,
David.

I get what yer saying and yes I understand its not magic......LOL I was just extremely happy with the product and figured i would share my thoughts on it. you guys are the experts here and thats why I come here. So if I offended some one i do apologize.

I can't speak for others, but I did not find anything you've said as offensive. Regarding the use of the term magic, it was my presumption and I apologize for making it.

If you're interested, I've figured out a way to perform a solo ABX an mp3gained file and a post-amplified version of the same file without the application of mp3gain. I hope kjoonlee will approve of my method.

• Make an mp3 from a wave file and use mp3gain to determine its find its max no-clip gain.
• Delete the mp3 file and apply attenuation to the wave file so that there is at least 6dB of headroom above what is necessary to prevent the mp3 from clipping (the max no-clip gain).
• Make an mp3 from this attenuated wave file and make a second copy of this mp3.
• Apply 6dB of gain to one of the mp3s using mp3gain and remove the tag containing replaygain and undo data.
• Load this file up in foobar2000 and manually apply a replaygain track value of 0dB.
• Load the other file in foobar2000 and manually apply a replaygain track value of 6dB.
• ABX the two tracks while telling foobar2000 to apply replaygain.

I think I speak for everyone here when I say that we're very happy that you're happy (that's what it's all about, after all), and that we appreciate your opinion.

And if I don't speak for everyone -- well, you get the idea.

Sounds like a plan and thanks for the advice.

Hey, no worries! You certainly didn't offend. I'm glad you find it useful.

Cheers,
David.

I have included clips for comparison...
http://www.hydrogenaudio.org/forums/index....showtopic=57614

Edit: OMG, all your proposed tests are worthless in this particular topic! Forget everything about foobar and other players. These relies on very high quality MP3 decoders, but you forget that the topic starter are using a car stereo. The only fair test would be to have a MP3 gained track and a ungained track which are played back with the same volume reduction ON THE HEADUNIT!


I can confirm this is a serious issue with my own Pioneer car headunit. It's poor mp3 decoder made my 89db MP3Gained mp3's sound awful.

My solution would be to make a foocar2000 in the future

Not really. hardrockin80s burnt them for listening.

Ah, yes, but for you to trust what your ears are telling you, you need to remove other possibilities from the equation. If you know which one is which, your mind can influence which one you think sounds better. The point of an ABX test is that its like the "Pepsi Challenge:" you don't know which one is which so subjective biases get removed from the equation to some extent. No one's trying to make you feel retarded, they're simply pointing out that if you want one thing to sound better than another (or even simply expect one to sound better than another), and you know which is which when comparing, you are quite likely to believe that the one you wanted/expected to sound better does sound better. A blind test limits the impact of expectation bias on that result. HA takes this point very seriously because so much of the "audiophile" world doesn't.

Another way to look at it is, as you said, you're happy with it so don't worry about it. It's your music, after all.

Where did he state that he burned them as audio CD's afterwards instead of just playing them as mp3?

Note the dates, please.

Where did he state that he played them as mp3?

You can't burn an mp3, or what are you reffering to??


That would be obvious to assume since he MP3Gains them instead of WAVGain them...

Not really.

I'm getting confused - what are you guys trying to prove here? That mp3gain=a volume control and nothing else? You can do that without ABX - just compare sample values objectively.

If you're trying to prove that songs at different levels sound different through a given stereo...?!


btw, the methodology, if I understand it, isn't fair.

mp3 encode > mp3gain
does not equal
wavegain > mp3 encode

...for various reasons. Accuracy (1.5dB units vs unlimited), dither/lack of, and most importantly certain mp3 encoders being level dependent (somewhat - depends on version too) so the two mp3 encodes are not equivalent.

Cheers,
David.