Why does RG compute the album gain differently when subfolders are included? I made an experiment in which I copied two albums (each in its folder) in the same folder, then replaygained them and compared the results to the same two albums replaygained separately, one after the other. Album gain differed, title gain was the same.

What should I do if I want to RG my music collection? Do it as a whole, including sub-directories, or folder after folder?

Would using replaygain with sweep be the same as gaining each album separately? How would the commandline read? I tried to find it out but wasn't successful.


greets
theduke

Question 1: WHICH ReplayGain?

Most likely, whatever you are using assumes each folder is one album, so if you copy both in a single one, it's not going to work correctly.

I assume that you're talking about mpc replaygain?

1) I won't get into technical details, but here's how I see it:

a=albumGain(1) = average (gain of files in (Album1)
b=albumGain(2) = average (gain of files in (Album2)

c=albumGain(1,2) = average (gain of files in (Album1, Album2))

I hope you realize that a,b,c aren't and shouldn't be equal, because they are based on different input data to begin with.

2) The replaygaining should be done album by Album.

PS. the tech & programmer guys might get upset with the 1) explanation because of the way I wrote it

my fault, replaygain for MPC.

I didn't copy all files simply into a single folder but I copied both album folders in one folder.

I know and I do hope that a, b, c are not equal. What I was referring to is the fact that the album gain value differs when

y) I copy different album folders in one folder and do replaygain including sub-directories
z) I replaygain each album separately

step 1: RG album folder a
step 2: RG album folder b

vs.

RG folder c including sub-directories (where album folder a and album folder b are sub-directories of folder c)

result: album gain for album folder a && b computed like in step 1 != album gain for album folder a && b computed like mentioned after vs.


ad 1) I thought the same. That's what I was wondering about.
ad 2) Therefore I wanted to know if I should do RG including sub-directories or rather RG each album separately.

What still bugs me is why the album gain differs. Because the different album gain values are not the same, depending on how they are computed, it makes me wonder what RG takes into account in album mode when including sub-dirs. In track mode it approaches all tracks to a certain level I think. But why not simply do this for a whole album then (besides maintaining dynamics differences of an album)?

Yes.




- Make sure the necessary utilities (sweep, replaygain, mppdec) are in the PATH.
- Open a DOS box and go to the folder in which your albums reside, each in its own folder.
- Issue the command "sweep replaygain --auto *.mpc".




It's exactly like Garf said. The MPC ReplayGain utility isn't advanced enough (yet) to treat each input folder as an album - it will assume that all input files belong to the same album, which is what happened in your case. Currently, using sweep.exe or foobar2000 (the latter requiring correctly written APE tags per album) is the only solution to work around this limitation.

Thx, now I understand. I thought replaygain would be smarter and thought Garf's answer only applies if I copy all tracks into a single folder.