Some of you sure will have noticed it: MPC-ReplayGain V0.84 does not calculate the album-values correctly, when you do it by "mass-replaygaining" (e.g. with MPC-batch-encoder). In that case, it sees all songs of all directories (= albums) simply as one album, and recalculates the wrong values.
Could that be fixed like it is in the ogg-vorbis replaygain-routine? There every directory is seen as an album, and the values to it are calculated correctly.

behold.... the genius has spoken...

Someone correct me if I'm wrong, but I believe that using ReplayGain with Sweep should take care of your problem.

The usage would go something like this:

sweep replaygain --auto *.mpc

That should auto-calc the track and album gain for every .mpc in all the subdirectories, and count each individual directory as an album.

Unfortunately, I can't locate this file... but I'll put up the binary... here:
http://www.knezek.com/Stuff/Sweep.exe

I know "Sweep" and as far as I know (and correct me if I´m wrong) sweep does exactly the same than the MPC-batch-encoder does. - So.... wrong values.

MTRH: Is this you or really your cat on the photo?

I am not aware of such a problem with replaygain, but if you are concerned just use foobar2k to rgain your mpcs. Works fine for multiple albums.

I applied ReplayGain to ninety or so albums a few months ago with Sweep.exe. Worked exactly as it should: files in the same directory are considered one album.

Okay, thanks Trelane to clear this up. So maybe I will jump back to sweep.exe again, as I used it already more than half a year ago. Then I moved over to the MPC-Batch-Encoder, because I hate those simple, non-comfortable commandline-tools.

I made some rg-tests a few minutes ago....
Result is as follows: While replaygain used with sweep brings correct values, the MPC-batch-encoder generates wrong album-gain-values, as it handles all gained directories only as one. So it´s not the fault of replaygain v0.84. -> You can forget this threat.

http://www.saunalahti.fi/~cse/files/Sweep.zip