Re: It happened again.

On Mar 13, 9:43 pm, "MrT" <symbiotr...@xxxxxxxxx> wrote:
On Mar 13, 5:36 pm, "david7ga...@xxxxxxx" <david7ga...@xxxxxxx> wrote:

The other night Grace Note identified the second CD of
Karajan's [...]

I had a recent similar experience. It is not a reliable system.
[...]

None of this is in any way remarkable. How iTunes and CDDB work has
been discussed previoulsy on r.m.c.r. You can find detailed
explanations of the original mechanism here

<http://en.wikipedia.org/wiki/CDDB>

and here

<http://ftp.freedb.org/pub/freedb/misc/freedb_howto1.07.zip>

The original discID was a good approach, but it was not a unique
identifier. "Collisions" (different audio CDs with the same discID)
did occur, more frequently the fewer the number of tracks on disc.

Gracenote does not disclose precise information about its database
(the commercial successor to the original CDDB) and the CDDB2
protocol; licensees, of course, are bound by NDAs. However, something
may be gleaned by examining the results -- for instance, an AIFF
ripped with iTunes/Mac from an unknown audio CD. AIFFs (like WAVEs)
are not just raw audio, and the additional data may include audio CD
layout info (length, track frame offsets), including a discID
identical to the discID calculated by the original method. After
iTunes contacts Gracenote, this number is changed to a new one in a
different format. Conclusions are obvious.

Both freedb and Gracenote allow for a certain "fuzziness" when
searching their databases for a match to an audio CD (i.e., the match
doesn't have to be perfect), but the extent of this fuzziness varies.
Again, it's one of those issues which Gracenote does not make public.

So it doesn't matter whether an audio CD is a re-issue of a recording
originally released on LP, or a home-made LP-to-CD transfer -- as long
as the track number and lengths match close enough to a record in
Gracenote's database, iTunes will assign the tags in that record to
the CD. The data in the original CDDB was user-contributed, and
consequently contained a significant number of errors; the extent to
which Gracenote's database is still user-contributed is, yet again,
confidential. Gracenote does employ editors for the database content,
but there's no way of knowing what and how much they actually do. It's
reasonable to assume they concentrate on pop music, since, business-
wise, that's where the Schwerpunkt is. I personally have not noticed a
difference in quality in data retrieved from Gracenote relative to
data retrieved from freedb, but that may be due to my musical
preferences. Classical music tags, in particular, are prone to error,
partly due also to the fact that CDDB's original design did not take
into account the more complex needs of this "genre".

Note that CD-Text

<http://en.wikipedia.org/wiki/CD-TEXT>

has nothing to do with it. As it happens, most CDs I listen to do not
contain CD-Text (which, again, may have something to do with my
preferences). I wonder if labels have general policies on CD-Text, or
decide to use it or not on a case-by-case basis.

Unfortunately, ISRC

<http://en.wikipedia.org/wiki/Isrc>

also has nothing to do with it. It's an established international
standard and it could have been for audio CDs what ISBN is for books,
but it's more often ignored than used, so it's of almost no practical
value.

There are other disc identification systems, but they're not used by
iTunes. Recently, Gracenote has announced CMI (Classical Music
Initiative), but released few specific details of what appears to be
no more than an enhanced tagging method.

The only noteworthy question in all of this is why was the iTunes/
Gracenote issue blown out of all proportion in the recent Hattogate
coverage. The answer is probably that most of those who wrote or
talked about it had no clue what iTunes did; that critics and
reviewers were as ignorant of technical details as they were lacking
in critical judgement; and that journalists were as lacking in
investigative spirit as they were quick to parrot each other. To my
knowledge, the first cogent explanation of this point in a major news
source was an article in the French-language "Liberation"

<http://www.liberation.fr/transversales/grandsangles/237539.FR.php>

published two weeks into the scandal.

--Alex


.

Relevant Pages

  • Re: iTunes for classical recordings
    ... Gracenote, which used to be called CDDB. ... iTunes uses Gracenote, and so every time I input a CD into it I have to do ... There is another database, this one evidently public-domain the way CDDB used ...
    (rec.music.classical.recordings)
  • Re: MP3 ripper from CD besides iTunes?
    ... populated by Gracenote via iTunes and not whatever database Max now uses. ... Files are not added to iTunes' library. ... I will reinstall Max and try it out. ...
    (comp.sys.mac.apps)
  • Re: Hatto - track durations for Liszt Studies anyone?
    ... I ran a few tests regarding iTunes' internal ... database, Gracenote CDDB, and freedb some time ago. ... If it was greater but <0.9%, then Gracenote and iTunes ... but no entry for Simon without Hatto or mentioning the BIS label. ...
    (rec.music.classical.recordings)
  • Re: iTunes for classical recordings
    ... Gracenote, which used to be called CDDB. ... CDDB, and how classical-unfriendly I have found it. ... Fortunately, iTunes (and some ... EAC does not support the ID3 "Composer" tag. ...
    (rec.music.classical.recordings)
  • Re: ipod itune songs synchronization in different machine?
    ... my choosing in my house and play audio through my cars' stereo systems. ... without EVER having to get iTunes involved, or even having iTunes on my ... well designed, fully featured media cataloging app -- iView Media Pro, ...
    (comp.sys.mac.apps)