grip/cdparanoia/freedb disagree on track length

Jamie Zawinski <[email protected]> Thu, 23 Jan 2003 18:24:52 -0800
Newsgroups gmane.comp.audio.freedb.devel,gmane.comp.audio.cd-paranoia.devel
Organization my own bad self
Message-ID <[email protected]>
I have a CD where cdparanoia and Grip / FreeDB disagree on the length
of the last audio track.  Grip and cdparanoia think the track is of
length 7:06, but the FreeDB data seems to say the track is 9:38.

The CD in question is "Phi in the Sky" by Kidneythieves.
It has 6 tracks, and tracks 1-5 are fine / consistent.

The FreeDB data says that track 6 is supposed to be of length 9:38.
When Grip generates a cddb file itself, it also says 9:38 (that is,
Grip generates cddb files consistent with the one that is currently
on the FreeDB servers.)

The Grip display, and "cdparanoia -vsQ", say that track 6 is of length
7:06.  That's also the length of the file I get when I rip it, with
either grip or cdparanoia.

Grip and FreeDB think that there is a 7th track, a data track.
The grip display shows:

    01  5:01
    02  6:26
    03  3:52
    04  8:57
    05  5:15
    06  7:06
    07  0:06  (data track, presumably?)

    That adds up to 36:43, or 2203 seconds.

cdparanoia -vsQ is in agreement, except that it doesn't list track 7.
It has this to say:

    Table of contents (audio tracks only):
    track        length               begin        copy pre ch
    ===========================================================
      1.    22635 [05:01.60]        0 [00:00.00]    no   no  2
      2.    28902 [06:25.27]    22635 [05:01.60]    no   no  2
      3.    17393 [03:51.68]    51537 [11:27.12]    no   no  2
      4.    40317 [08:57.42]    68930 [15:19.05]    no   no  2
      5.    23625 [05:15.00]   109247 [24:16.47]    no   no  2
      6.    31968 [07:06.18]   132872 [29:31.47]    no   no  2
    TOTAL  164840 [36:37.65]    (audio only)

Note that the total length is off by 6 seconds.  I think that this
is because the unshown data track is 6 seconds long (see below.)

The freedb data is here: http://www.freedb.org/freedb/rock/5a093307
Converted to seconds, the numbers in that file come out to:

    01  5:02
    02  6:25
    03  3:52
    04  8:58
    05  5:15
    06  9:38

    That adds up to 39:10, or 2350 seconds.  This implies that
    the data track is 0:06 long (since the cddb file says the
    total file length is 2357.)

My standalone CD player (Nakamichi MB-10) is also in agreement: it
says there are 6 tracks (not 7!), total length 36:38 (2198) with
track 6 being 7:06 long.

So how am I ending up with CDDB / FreeDB figures that claim that
track 6 is 9:38 instead of 7:06?

Might the TOC have overlapping tracks, or something?  Is this some
new sleazy way of hiding data tracks from audio CD players?

The reason I'm concerned about this is that I have a script that
issue warnings when my MP3 files are not the length that the CDDB
files say they should be, so that I can tell when I've somehow
mis-ripped something.  That script is warning about this track,
and so I either need to know how to rip the track properly, or
how to interpret the numbers in the CDDB file to realize that
the track *was* ripped properly.  Because right now, the CDDB file
is claiming that the file is two and a half minutes too short.

Any suggestions appreciated...

Versions:

    cdparanoia-alpha9.8-11
    grip-3.0.1-4
    Red Hat Linux release 8.0 (Psyche)
    Linux 2.4.18-14 #1 Wed Sep 4 12:13:11 EDT 2002 i686
       athlon i386 GNU/Linux

-- 
Jamie Zawinski
[email protected]             http://www.jwz.org/
[email protected]       http://www.dnalounge.com/