Re: [Gtkpod-devel] Re: [PATCH] Reading LAME ReplayGain Values

[Jens T. <jen...@rw...>]

On Mon, Mar 15, 2004 at 06:21:29PM +0100, Jens Taprogge wrote:
> Hi Jorg,
> 
> Thanks for putting this into CVS. 
> 
> (btw: is there a way in cvs to do automatic incremental patches? Like:
> generate a patch with 
> # cvs diff -u >patch1 
> and then gerate a second one on top of that
> # cvs diff -u --somefancyoption patch1 > patch2
> I did not find anything in the docs. )
> 
> I am also thinking about implementing APETAG ReplayGain reading (this is
> where mp3gain stores the information. I think it would be much cleaner
> than parsing the output of mp3gain. Also the external application would
> not have to be rerun also the time if the gain already has been
> calulated. And users that don't have mp3gain installed could still use
> the information provided. It does not seem to complicated.
> 
> One thing I notices while working on the last patch: There are two new
> options in the prefs (the datatype) but only one in the prefs_window. It
> took me a while to figure that out. Is it intended? Also I think the
> names are somewhat to similar for my taste.
> 
> Regarding AlbumGain / AudiophileGain: One thing I thought of was
> implementing a genre based automatic offset. To for example make a Jazz
> mp3 somewhat more quiet compared to for example a Hard Rock tune. With
> all the songs normalized to one volume it is kind of weired listening to
> songs of different genres after one another. I think that combined with
> AlbumGain should give a pretty good estimate of the 'correct' volume.
> What do you think?
> 
> Later
> Jens

I have moved the volume calculation as discussed. Also I have removed
the automatic setting of the Volume in mp3file.c. I think it should gp
somewhere else and probably by on option ("set volume on import").

Apart from that I have implemented APE Tag reading.
I would suggest the following procedure: 
  When the user tells the program to read calculate the volume
  normalization we first try to see if there is a replay gain set in the
  database (this would come either from the LAME or the APE Tag - in
  that order). If not we could try to reread the file in case the
  database is from an old gtkpod version or the file has been updated.
  We can then check the database again. If we are still unsucessful we
  can run mp3gain. And reread the file again. No need for nasty parsing.
  
  On the other hand that would require mpp3gainto always write to the
  file. Not sure if that is acceptable...

Btw. I can not seem to make GtkPod update the volume in the display for
some reason. The database is ok.

Later 
Jens

-- 
Jens Taprogge
mailto:jen...@rw...