Menu

#51 Mp3 Volume Normalization: (mp3gain ?)

closed
1
2006-06-23
2006-06-14
Vathek
No

Hi,
is it possible to include in ml_ipod a feature that
normalize all the volume levels of songs on ipod?

Normalization consists on analyze the volume of songs
on the ipod and making them at the same peak level...
(so that switching from Vivaldi to Metallica could be
possible without constantly modifying volume wheel)

Actually I'm using mp3gain
(http://mp3gain.sourceforge.net/) to scan all file on
my ipod and then normalize the volumes...

Mp3Gain works by adding some information tags about the
original and wanted (read normalized) volume level to
the mp3s (this is the first step called "analyze").

Then the second step is to actually normalize (changing
the mp3 volume) the songs... (called "Normalize" feature)

All this method is completely tunable and reversible as
normalization keeps the original volume level tag so
that is always possible to undo everything...

In my opinion it could be possibile to add such a
feature in "Tools" or maybe just before the transfer...

Anyway it seems to me to be not so different from
transcoding (already implement) so ... give it a look :)

Thank you :)
(Mp3Gain was also supported by EuPod, because is also
avaiable a nice mp3gain command line executable)

Discussion

  • Vathek

    Vathek - 2006-06-14
    • summary: Mp3 Volume Normalization Request... --> Mp3 Volume Normalization: (mp3gain ?)
     
  • Achim Bursian

    Achim Bursian - 2006-06-16

    Logged In: YES
    user_id=466564

    Just to make it clearer: You would like to have it that way:

    When a file is transfered to the iPod, it should be copied
    to a temporary place first, then mp3gain exe should edit
    it, and then we write the edited file to the iPod. Correct?

     
  • Achim Bursian

    Achim Bursian - 2006-06-16
    • priority: 5 --> 1
     
  • Vathek

    Vathek - 2006-06-16

    Logged In: YES
    user_id=1539641

    Ok, normalization could be done either before or after
    transfer .

    As normalization step requires the analysis one, and two
    steps require about 10second for track, I think the best way
    is to add a command that do the normalization directly on
    the ipod...

    (Obviously normalization should be done only on new songs,
    but I think that this task is already done by mp3gain itself)

    Hope it's more clear now :)
    Thanks!

     
  • Achim Bursian

    Achim Bursian - 2006-06-19

    Logged In: YES
    user_id=466564

    Ok, thanks for the clarification.
    I have a new test version up on the website. It does
    mp3gain on all songs (There is an option in the tools menu)

    No prefs page yet, it simply uses the standards. And
    mp3gain must be correctly installed (with the installer).

    On my nano, it takes ages, I guess this is because of the
    slow read/write speed on the flash memory. Could you give
    it a try on your iPod (which model do you have?)

    Is that ok this way?

     
  • Achim Bursian

    Achim Bursian - 2006-06-19
    • assigned_to: nobody --> achim66
    • status: open --> pending
     
  • Vathek

    Vathek - 2006-06-19

    Logged In: YES
    user_id=1539641

    I have a 4MB Mini version... I'll test your new version
    immediatly..

    Anyway did you try to test processing time by running
    mp3gain on a dozen of tracks either on hard disk and on your
    nano flash?

    High timing could be due to a very slow cpu...
    I'm using a P4 3,2Ghz and it takes approximatively the same
    time on my mini and on my hard disk)

     
  • Vathek

    Vathek - 2006-06-19
    • status: pending --> open
     
  • Vathek

    Vathek - 2006-06-19

    Logged In: YES
    user_id=1539641

    OK! It works good with my ipod mini.
    Actually I ran it only for normalyzing task (as I already
    did the analysis with mp3gain original program) and it takes
    15-20 tracks per minute, so it's quite the same speed as the
    original mp3gain.

    Anyway I also found some strange little flaws:

    1) I found that it normalizes all tracks to 91dB. It's
    strange since mp3gain default value should be 89dB? (Is it
    intentional or it's a problem of mine?)

    2) I found that tracks of 15minutes are not being
    normalized, I don't really know why as original mp3gain
    normalises them without any problem...

    Hope it helps :)

     
  • Achim Bursian

    Achim Bursian - 2006-06-20

    Logged In: YES
    user_id=466564

    Thanks for reporting that it's working fine.
    The two flaws are actually implemented that way: The 91dB
    just to see if other dB settings work, and the 15min limit
    that very long tracks (audiobooks) get excluded. 15min is
    hardcoded at the moment. That will go into a prefs page.
    Don't you think that this limit is a good idea?

     
  • Vathek

    Vathek - 2006-06-20

    Logged In: YES
    user_id=1539641

    Ok, preference page it's a good idea.
    Just include dB and time limit in the user settings and the
    feature will be perfect. :)

    For default values I think 89dB (maybe 90?) and 15min are ok!

    Thanks!

     
  • Vathek

    Vathek - 2006-06-23

    Logged In: YES
    user_id=1539641

    I've just tried v1.40p10 and it's look very very good...

    However at this point I'm thinking about a pair of
    additional features that could make it better:
    1) Is it possibile to show in winamp ipod plugin the volumes
    tag as it happens with original mp3gain? (In the column near
    the id3 tags as size or bitrate for example)
    Mp3Gain tags are in APEv2 format.

    2) While normalizing I can't know how much time process will
    require, is it possible to add an estimation of time needed
    to complete? (As it happens with transfer process?)

    Thanks a lot, you're making an excellent plugin! (Maybe
    better than iTunes itself!)

     
  • Achim Bursian

    Achim Bursian - 2006-06-23

    Logged In: YES
    user_id=466564

    Ok, to your questions:
    1) No, ml_ipod does not read ANY tags from the file. That
    would be too much work to add. At the moment, it only
    displays what it gets from the ML interface, and there are
    no APE2 tags available.
    2)No again. I don't know in advance how long it will take,
    only how many files I have to process. For files that have
    been procesed already, it takes milliseconds until mp3gain
    returns control. But for those that have to be analyzed or
    changed, it takes several seconds. And I leave it to
    mp3gain to detect which ones it has to analyze.

    To support what you request, it would be necessary to scan
    all files in advance to see whether processing is
    necessary. That takes time again. I think it's not worth
    the effort.

     
  • Achim Bursian

    Achim Bursian - 2006-06-23
    • status: open --> closed
     
  • Vathek

    Vathek - 2006-06-23

    Logged In: YES
    user_id=1539641

    Ok for the first topic, but for the second one I think it
    could be possible to give an estimation based on the mean
    value of processing time for a track.

    I mean that just after have processed one track I have a
    mean value identical to processing time for that track.
    Then after two tracks I will take half of the total time for
    mean processing time.
    So after N tracks I will take total time elapsed divided by N...

    At this point if there are T-N tracks remaining I always
    could estimate the remaining time applying mean value to the
    remaining tracks.
    So a simple formula to do this is:

    (TOTAL_TRACKS-N)*(TIME_ELAPSED/N)
    where N is the number of tracks allready normalized.

    Simple, isnt'it? :)

     
  • Achim Bursian

    Achim Bursian - 2006-06-23

    Logged In: YES
    user_id=466564

    You were right if there was something like an average time.
    But imagine: You have 5000 tracks, all normalized already.
    Now you upload 100 new ones, which will take average 8
    seconds per track. The 5000 first ones will be analyzed in
    a second or so, so the calculation will always
    say "estimated time left 1sek". Even after normalizing some
    of the new files, the average time will be very small, so
    no real estimation can be done.
    And I don't want to keep track of what has been normalized
    already, besides the tags that mp3gain writes.

    Surely it would be possible to do something about that, but
    as you know, I do this in my spare time only. And I've got
    a regular job, that takes most time, too. So from my point
    of view, it would be just too much effort for something
    that is only "nice to have", but definitely no improved
    functionality.

    I waste too much time for ml_ipod anyway... my family needs
    more of it.

     
  • Vathek

    Vathek - 2006-06-23

    Logged In: YES
    user_id=1539641

    Ok, ok... no problem...

     

Log in to post a comment.

MongoDB Logo MongoDB