hello,

when I change the volume of my mp3 files I usually go with 95dB - 97dB...most songs still don't clip (at least they are not shown red in the list or something). Why is the default value then 89dB? I mean, even after I rip a CD the volume of the mp3 is ~96dB...

From
How to use Mp3 Gain

hmm

OK, first of all, I do not use MP3Gain.
Second: If I would use it, I would ignore the "clipping"-thing and also would chose something about 96dB. Otherwise everything is to quite. If you ever make VAVs out of these files, you only use 15bit out of 16 (6dB=1bit).

heh... ignore this "clipping" thing...

alright... here is the deal with MP3Gain... there are two reasons people use it:

1. Prevent clipping. It has been shown that compressing an audio file that did not clip with a lossy encoder will in some cases cause clipping. This is because of the way the audio is stored in the file. This can be fixed by setting the "volume" the file to a lower value, so that during playback, the decompressed stream never reaches full scale. This gain changing is a lossless procedure.

2. Equalize the volume of your music. It should be obvious that "97dB" in MP3Gain does not mean 97dB coming out of your speakers. If your music is too low, reach for the volume control. The number in MP3Gain is only there for reference. It means that if you change all the files to be, say, 89dB, they will all be at the same volume. Period. It has NOTHING to do with the actual playback volume, since there are at least three volume controls on the way from MP3 to your years (Winamp's, Windows Mixer and Analog Volume).

That dB number there is referenced to the fullscale "volume" of a theoretical sine that will be as loud as it can be without clipping. I don't remember the value off hand. If you've used MP3Gain with a few songs, you will notice that some of them start clipping when normalized to 97dB, some I've seen as low as 90dB. 89dB was chosen because it is low enough that any file normalized to that volume will not clip. So you can let it run thru your whole MP3 collection without giving it a second thought.

If you do that, and use radio gain, then all of your songs will have the same perceived volume. That means that you can start playback, set your volume knob to whatever, then sit back and not have to touch it again. There will be no songs too low for you to hear, or songs too high for you to be annoyed with. Of course, this is on the average.

If you use album mode, then all CDs will average out to the same volume, but each song on the CD will be louder or quieter in proportion to their original album volume.

Yes, but tell me if you prefer to hear clicking over a 1 bit fidelity loss.

If you do, man, you are deaf.

the problem might be that the weak amps in portables/cars already expect the source to be around the 100 dB area, and dont have the power to playback 89 dB files really loud even if you turn the voulme knob up to 100%. Thats not replaygain's fault... test your portable/radio with a replaygained CD before you buy it

Well, MP3Gain tells me, that 90% of my APS LAME-Encoded Tracks have clippings. I never had noticed one. Do you notice them? So then you should better give this info to the developers of LAME, so they can react on this problem.

If you are manually setting your mp3's at that db range you should read about the maximizing feature and why it shouldn't be used.

I believe although cannot confirm at this moment since I don't have MP3Gain installed that there is an internal link within the program that will take you to a webpage detailing about the maximizing feature 'max no-clip gain' which is what I think you are essentially applying to your mp3's.

In threads such as this one, two basic camps of people arise.

They both rip, and Analyse in MP3Gain, then...

Camp 1) None of the files are clipping, and they can either boost up to about 95db, and still get no clipping.
Camp 2) 90% of the files clip when they are ripped.

I fall into the second category.

Is this just down to differing music tastes, or is the cosmic joker at play here?


[ Ok, well maybe there hasnt been other thread(s) that say this, but I have it in the back of my mind that there has been ]

Aren't they the same camp? If the CD is a modern pop recording, it will hit -0.1dB FS (i.e. a fraction below digital full scale) very regularly. This means it will probably clip when it's mp3 encoded.

This also means that the file will
a) be very loud
b) have very little dynamic range

As a result, ReplayGain will try to make it much quieter. It will have to if you want it to be 89dB average loudness - that's 6dB quieter than a typical over-compressed pop song (but still 6dB louder than the normal speech in a movie sound track).

After ReplayGain-ing to 89dB, there will be a huge amount of headroom between the peaks of the recording and digital full scale, hence the target volume can be increased without introducing clipping - until you get near to the 95dB the track started at!


There was once some confusion between "it will clip if replaygained" and "it already clips", but I think that's clearer in recent version of mp3gain.


Cheers,
David.


edit: smiley -> b)

Yep, you are right; my post is pretty cack

[Note to self: Engage brain before posting ]

Nah - I only knew the answer because I'd thought exactly the same thing, and then had to think it through.



Where would be the fun in that?

so in conclusion, we should leave the target volume as 89dB?

yup