(I'll send a copy to gtkpod-devel -- please subscribe and answer to there)
On Sat, Aug 09, 2003 at 12:53:00AM +0200, Edward Matteucci wrote:
> If we want to add normalization to gtkpod we should answer this question:
> how iTunes normalize the mp3 files? Is the iPod "normalization
> aware"? Does it adjust the sound volume in some way?
There's a volume adjust field in the iTunesDB for each track. I have
added support for this field (new song attribute column) and will
upload the code to CVS now. All you have to do is to set this field to
a value between -100 and +100 to adjust the volume of the
> I have found that there are two ways to the normalization:
> detect volume level-->decode (mp3->wav) file-->normalize wav --> encode (wav->mp3)
Not necessary with the iPod.
> detect volume level--> add a metadata keeping the sound level
That's the way to go. "Metadata" simply is the newly supported field.
> the first strategies can be found pretty everywhere but it's:
> time consuming
> cpu&hd intensive
> not very "sound quality" friendly
> not usable on a big mp3 collection (10Gb are a big collection)
> the second one is pretty fast but we should understand how the iPod
> want this tag.
> I've seen a PowerBook normalize my mp3 collection on the iPod (1500 files).
> it was a long operation (around 25 minutes) but it's not even
> comparable to decoding&encoding operation.
> I think that they use something similar to mp3gain based on the
> replaygain proposal.
We could try to use the mp3gain code (as you pointed out, it's under
LPGL). mp3gain is not well supported under Linux (took some effort to
compile 1.4 -- it obviously was never tested) -- so it would be
difficult to ask people to just get it as an external program (like we
do with xmms). mp3gain is slow, but quite promising.