Dynamic range compression tool...

[James Gray]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Paulo Matafome Oleiro wrote:
> Seg, 2006-07-24 às 18:19 +0200, Andrea Giuliano escreveu:
>> mp3 audio file (or wav file, maybe) so that the difference in
>> loudness between very loud passages and very low passages 
> Sincerely
>
> The only program that I know is MP3 ReplayGain (mp3gain) to get all
> the track louder.
> This should be available at the repositoires

The only thing this program does is read the entire MP3, find the peak
value, and bumps the total gain up or down so the peak matches whatever
you want it to be.  It does this by adding some meta data to the
beginning of the MP3.  It doesn't actually "recode" the file as such.
This might be exactly what the OP was after, although from personal
experience, peak value is of limited value.

I usually encode my MP3 from the CD's and "normalise" them at the same
time.  This method uses the average gain for the entire track and fudges
the gain based on that.  Imagine the difference between the perceived
sound of a mellow classical piece with a big crescendo finale being
"peak analysed" as opposed to averaged - the bulk of it would be
quiet-as-a-mouse, whereas the averaging method will lift the bulk while
letting the big finale blow your eardrums :P  Just what the composer wanted!

Then there's other methods that use a sliding average based on the gain
in adjacent n-frames, rather than averaging the entire track.  This is
the most "accurate" method but can also alter the "feel" of piece of
music.  It's also the most CPU intensive as each frame is recoded based
on it's adjacent frames.[1]

Different methods suit different styles, so like I said, mp3gain might
be exactly the right tool depending on what you're trying to do.

Cheers,

James

[1] Don't quote me on that - my recent track record on lossy format
explanations hasn't been all that good[2].  But you get the general
idea.  I'm not even sure the sliding average method can be used on
existing MP3's without loosing quality or whether it has to be done
during the encoding from the original source (to avoid losses beyond the
base bit rate etc).

[2] In this case, I'm probing dark recesses of brain that were written
during a class at University many years ago.  I remember the gist of it,
but not the details.
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (Darwin)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFExd66wBHpdJO7b9ERAqRSAKDW8XR19IXocx7sPRP6rQmRWYpm5QCgqd++
LRxd4+5mtO1lRZ835NpX4bA=
=ZJQ1
-----END PGP SIGNATURE-----