Hatto - track durations for Liszt Studies anyone?
Andrew Rose
question. Gracenotes reckon that the Hatto track durations shouldn't
prevent it from identifying the CD as by Laszlo Simon, even though
they're quite different in some cases.
I'd like to test this out by putting together a number of dummy CDs,
varying the track lengths, and seeing how far you can push Gracenotes
before it stops making these kind of identifications. Was it really
iTunes which started this off, or was the information planted there for
someone to find?
If someone has an accurate (i.e. from the Hatto CD, not the cover notes)
list of track timings I'd be most interested to see them - for the Hatto
Liszt Transcendental Studies CD.
--
Andrew Rose - Pristine Classical
The online home of Classical Music: www.pristineclassical.com
David Fox
Yeah, the "iTunes identification" story really doesn't hold water. I
thought that from the second I read it. There are two possibilities I
can think of:
1. Somebody concocted this story to cover the true source of their
info. The source spilled to the press on the condition of anonymity.
It's dicey to print such an inflammatory story without named
attribution, true or otherwise. The "magic iTunes" story obviates
these issues. Besides, how many people really understand computer
algorithms anyway?
2. Somebody purposely entered the Simon data for the Hatto disc
hoping that somebody would notice.
1 is much more likely than 2 for no other reason that the timing and
probability of discovery in scenario 2 are so uncertain. To be
honest, I really don't care how the story broke. It doesn't change
the material issues of the Hatto Hoax.
My understanding of the mathematics of the Gracenote/CDDB algorithm is
that there is probability zero of an accidental match if the number of
tracks do not match between two selections regardless of the timing
matches (or near matches) of any of the individual selections. The
articles I've read suggest that the Simon and the Hatto releases do
not line up thus: Simon did not record all 12 etudes and "Joyce"
did. If that's the case then you can save yourself the work.
DF
Andrew Rose
> On Tue, 06 Mar 2007 21:58:32 +0100, Andrew Rose wrote:
>
>> Just one loose end that I'd like to try and tidy up - the iTunes
>> question. Gracenotes reckon that the Hatto track durations shouldn't
>> prevent it from identifying the CD as by Laszlo Simon, even though
>> they're quite different in some cases.
>
>> list of track timings I'd be most interested to see them - for the Hatto
>> Liszt Transcendental Studies CD.
>
> tracks you got from him should be quite accurate.
>
> In my experience the detection range could well be sufficient. I once
> digitized a recording from a LP. But, as the LP was a bit damaged, I took
> the casette recording I made about 25 years ago. At that time I used a
> Lenco L75 turntable with a duster (not really perfect for speed accuracy).
> I transferred the complete LP, from cassette, to a single WAV file (with a
> different cassette deck), added trackmarks and wrote a CD-RW. When I used
> this CD-RW for MP3 conversion, WinAmp correctly identified the LP/CD. I
> think the timing of the result hardly can be verry accurate, I would be
> surprized if they are accurate within 1%.
>
The different in the case of track five is about 12-13s. I just wonder
whether it's still going to pick up a CD with this kind of variance?
I can do further chasing up tomorrow, but in the meantime wondered if
anyone here had a copy of the CD...
MrT
have already mentioned. I suspect that given the mounting rumors about
different pianists playing in the different CDs, somebody decided to
snitch. Actually, the serious joke that Hattos were fakes had been
going around almost since the beginning of the shilling operation. I
think that should Hatto's recordings have been widely available, the
hoax would have been discovered quickly. As it was, most collectors
hadn't heard any Hattos, others had bought one or two, and so on. And
the people who allegedly had 90 Hattos available did not show much in
the ears department. Bryce Morrison? Ates Orga? Tom Deacon? David
Hurwitz? (actually, Hurwitz says he had not heard most ot his stash of
Hattos).
I think the easiest test for Gracenotes and iTunes would be the
Haebler recordings, which according to reliable sources are the same
as Hatto's. If you put Hatto and it doesn't suggest Haebler, something
is wrong. Likewise, if you put in Haebler and it doesn't suggest
Hatto. If the recognition doesn't work with recordings that are pretty
much exact copies... If it does, you can still do the Simon test, or
maybe you should do it anyway, since that's what is claimed to have
opened up the case. But I suspect it won't.
Best,
MrT
tomdeacon
> Andrew, I doubt the iTunes story for the same reasons other posters
> have already mentioned.
Baseless conclusion.
I suspect that given the mounting rumors about
> different pianists playing in the different CDs, somebody decided to
> snitch.
Unfounded suposition.
Actually, the serious joke that Hattos were fakes had been
> going around almost since the beginning of the shilling operation.
Doubtful gossip.
> I think that should Hatto's recordings have been widely available, the hoax would have been discovered quickly.
Incompetent surmise.
> As it was, most collectors hadn't heard any Hattos, others had bought one or two, and so on.
Most collectors?
Does that mean yourself?
> And the people who allegedly had 90 Hattos available did not show much in the ears department. Bryce Morrison? Ates Orga? Tom Deacon? David
> Hurwitz?
Hmmmmmm.
As opposed to MrT, David Fox, Taffy Brendel, Jackoff Hofmann?
(actually, Hurwitz says he had not heard most ot his stash of
Hattos).
And you believe Hurwitz?
> I think the easiest test for Gracenotes and iTunes would be the
> Haebler recordings, which according to reliable sources are the same
> as Hatto's.
Unsubstantiated statement.
> If you put Hatto and it doesn't suggest Haebler, something is wrong.
It doesn't.
Brilliant conclusion, though.
What's your IQ?
Likewise, if you put in Haebler and it doesn't suggest
> Hatto. If the recognition doesn't work with recordings that are pretty
> much exact copies...
Are the ... supposed to be a conclusion?
What is this, semaphore?
If it does, you can still do the Simon test, or
> maybe you should do it anyway, since that's what is claimed to have opened up the case. But I suspect it won't.
Baseless supposition.
Suggestion: get another job.
TD
Walter Traprock
> 2. Somebody purposely entered the Simon data for the Hatto disc
> hoping that somebody would notice.
I took it as obvious that this was the case! Wither their motivations,
who cares? It was a casual thing.
Anyways, here's the CD timings according to itunes:
1) :46
2) 2:16
3) 5:06
4) 7:53
5) 3:35
6) 5:22
7) 4:48
8) 5:19
9) 9:48
10) 4:20
11) 9:37
12) 5:29
12 songs, 1:04:19 total time, 650.2 MB
from the back of the case of CACD - 9084-2:
1) 0.48
2) 2.17
3) 5.05
4) 7.50
5) 3.44
6) 5.23
7) 4.48
8) 5.23
9) 9.45
10) 4.25
11) 9.40
12) 5.38
Overall Timing: 64.46
It's a great disc, I'm waiting for the old actual Laszlo Simon disc,
on Bis! The unmolested original probably sounds even better, and I
might like the Simon feux follets better than Nojima's!
"Recorded Digitally - The Concert Artist Studios
28th December 1990 & 6th January 2001"
at least they used a Steinway piano!
Walter Traprock
> Anyways, here's the CD timings according to itunes:
>
> 1) :46
0:46.626
> 2) 2:16
2:16.8
> 3) 5:06
5:06.213
> 4) 7:53
7:53.04
> 5) 3:35
3:35.72
> 6) 5:22
5:22.253
> 7) 4:48
4:48.12
> 8) 5:19
5:19.893
> 9) 9:48
9:48.986
> 10) 4:20
4:20.88
> 11) 9:37
9:37.08
> 12) 5:29
5:29.186
Andrew Rose
> Andrew, I doubt the iTunes story for the same reasons other posters
> have already mentioned. I suspect that given the mounting rumors about
> different pianists playing in the different CDs, somebody decided to
> snitch.
If you visit the Gracenote site you'll find the following, originally in
the Wall Street Journal:
http://www.gracenote.com/corporate/press/20070226WSJ.html
and in particular:
"...Mr. Rose's comparison of the Hatto and Simon CDs notes that track
times have been shrunk or stretched. The fifth track on the Hatto CD, he
says, is 13 seconds shorter than the fifth track on the Simon CD -- no
surprise, as Mr. Rose alleges the track is lifted from a Nojima
recording, not from the Simon CD. From classical-music newsgroups to
blogs and Slashdot, such discrepancies led some to doubt that Gracenote
could automatically match the Hatto CD with the Simon CD -- unless
someone had altered the Hatto CD's information in the Gracenote database
so that it came up as Mr. Simon's.
Mr. Rose spoke for many such doubters with this hypothesis: "Someone,
somewhere, used this as a means of bringing the possibility of a hoax to
the attention of the likes of the Gramophone critic. … It's almost
certain that someone deliberately submitted the Hatto CD information [to
Gracenote] with Laszlo Simon's name attached, and did this for a reason."
Sounds like James Bond stuff, but it isn't: Gracenote began with
information entered by CD listeners about their own music, and still
gets thousands of user submissions and edits a day.
So did the Hatto investigators benefit from a planted clue leading them
to compare Ms. Hatto with Mr. Simon? (Keep in mind that the initial
Gracenote identification has no bearing on the subsequent audio analysis
either way.) The idea seemed a delicious twist to a mystery that already
had twists a-plenty. But sorry, mystery fans – this one looks like a red
herring.
Ty Roberts, Gracenote's chief technology officer, says Gracenote does
allow for a certain amount of "fuzziness" in identifications: Remastered
CDs, or ones pressed at different plants, can yield tracks that are
slightly longer or shorter. Gracenote views fuzzy matching's exact
parameters as a trade secret, but Steve Scherf, the company's chief
architect, was able to knock down the idea that a clue had been planted,
at least to my satisfaction.
While warning that he hadn't had a chance to examine originals of the
CDs, Mr. Scherf offers this: "I can say with good assurance that our
system would have matched the apparently very similar Joyce Hatto and
Laszlo Simon discs. … With this many tracks, it is a near-mathematical
certainty that two discs are indeed the same if all tracks but one are
about the same length." In other words, an outlier like that fifth track
wouldn't be enough to change the mathematical conclusion that Hatto =
Simon. In addition, he says, a wholesale change like changing the artist
would have been flagged for scrutiny by Gracenote's staff."
It's something I've been asked about since by a number of journalists,
so I thought it best if I actually did some real testing of Gracenote,
rather than continue hypothesising....
Andrew Rose
Many thanks for this - I now have a dummy CD file which matches these
timings precisely. As soon as I get software installed which will
interrogate the Gracenote database (hopefully not iTunes if I can keep
it away from my system!) I can start investigating...
Kirk McElhearn
> Just one loose end that I'd like to try and tidy up - the iTunes
> question. Gracenotes reckon that the Hatto track durations shouldn't
> prevent it from identifying the CD as by Laszlo Simon, even though
> they're quite different in some cases.
I've said it before, and I'll say it again. This Gracenote thing is a
red herring. It did not identify the Simon Bis CD, but rather the
Concert Artists CD having as performer Simon. CQFD.
Kirk
--
Read my blog, Kirkville
http://www.mcelhearn.com
Kirk McElhearn
> So, did anyone check to see what happened if iTunes pulled data for the
> original BIS CD? If so, was it identified as a Concert Artist disc or
> a BIS disc, and when?
It's not in the Gracenote database.
Andrew Rose
>> In article <45ee564d$0$5085$ba4a...@news.orange.fr>, and...@pristineaudio.com says...
> <...>
>> While warning that he hadn't had a chance to examine originals of the
>> CDs, Mr. Scherf offers this: "I can say with good assurance that our
>> system would have matched the apparently very similar Joyce Hatto and
>> about the same length." In other words, an outlier like that fifth track
>> wouldn't be enough to change the mathematical conclusion that Hatto =
>> Simon. In addition, he says, a wholesale change like changing the artist
>> would have been flagged for scrutiny by Gracenote's staff."
>>
>>
>
> original BIS CD? If so, was it identified as a Concert Artist disc or
> a BIS disc, and when?
>
Well it looks like I might have to eat my words on the subject of
iTunes/Gracenote. However, the Hatto/Simon discs do seem to be on the
very limit of the Gracenote software's abilities.
I made 6 test discs and ran them through the Gracenote CDDB system:
CD1 - Timings identical to Hatto - offers Hatto or Simon
CD2 - Timings reduced on 3 tracks within Hatto parameters - offers Hatto
or Simon
CD3 - Timings reduced by same number of seconds as Hatto/Simon
differences to same tracks - fails to recognise CD
CD4 - Timings reduced by approx. 3/4 of Hatto/Simon time differences -
offers Hatto
CD5 - Timings reduced by percentage differences between Hatto/Simon -
offers Hatto
CD6 - Simon CD - offers Simon/Hatto
Really the last 2 point to the software doing the ID rather than someone
submitting the wrong name deliberately.
If you look at the number of seconds reduction - CD4 - you do fail to
find the disc in Gracenote. But because this is not a fully
proportionate reduction in track lengths, but very slightly greater than
the actual proportions involved, then this probably indicates pushing
Gracenote just past its limits.
Finally, the fact that the Simon disc also offers Hatto does suggest
that the software did manage to identify the 'Hatto' disc correctly,
without further intervention.
FWIW I used the Hatto data supplied here by Walter Taprock (the second,
more accurate post) and the timings supplied by eMusic (to the nearest
second). The differences are as follows, with Hatto always quicker where
the times have been changed:
Track 1: +2
Track 2: +1
Track 3: 0
Track 4: 0
Track 5: +13
Track 6: +1
Track 7: +1
Track 8: +4
Track 9: 0
Track 10: +7
Track 11: +2
Track 12: +15
Some of these differences will be due to tempo alterations, others to
changes in the gaps before and after the music, and in the case of
tracks 5 and 12, using different source recordings.
Finally, looking more closely at the timings of the Simon (to the
nearest 1/1000s) I wouldn't expect any marginal increase in accuracy
here from my original timings (rounded down to the nearest second) to
change these results.
Kimba W. Lion
>(hopefully not iTunes if I can keep it away from my system!)
Funny... that's exactly how I feel about it. So it's not just me.
Kirk McElhearn
> Well it looks like I might have to eat my words on the subject of
> iTunes/Gracenote. However, the Hatto/Simon discs do seem to be on the
> very limit of the Gracenote software's abilities.
Except for one thing: the Simon disc _was not in the Gracenote
database_ when this whole thing began. I checked several times, and
searched several ways for Laszlo Simon. He _only_ showed up as being on
the Concert Artists CD. So Bis - or Gracenote - has added the info
since then. Basically, so what you do returns the Bis recording, and to
cover their tracks for their tech person saying something ridiculous.
>
> I made 6 test discs and ran them through the Gracenote CDDB system:
>
> CD1 - Timings identical to Hatto - offers Hatto or Simon
>
> CD2 - Timings reduced on 3 tracks within Hatto parameters - offers
> Hatto or Simon
This suggests that they've also altered the database...
>
> CD3 - Timings reduced by same number of seconds as Hatto/Simon
> differences to same tracks - fails to recognise CD
So you're saying that the timings that correspond to the Hatto
recording don't return anything? For CD1 you are saying it does?
>
> CD4 - Timings reduced by approx. 3/4 of Hatto/Simon time differences -
> offers Hatto
>
> CD5 - Timings reduced by percentage differences between Hatto/Simon -
> offers Hatto
>
> CD6 - Simon CD - offers Simon/Hatto
These last three suggest that Gracenote has twiddled with the database
again. If the timings are as different as you suggest, they shouldn't
happen. You can try taking another CD - anything that's in the
Gracenote dabatase with a single record - then making a test CD with
slightly different timings to see what happens...
>
>
> Really the last 2 point to the software doing the ID rather than
> someone submitting the wrong name deliberately.
>
> If you look at the number of seconds reduction - CD4 - you do fail to
> find the disc in Gracenote. But because this is not a fully
> proportionate reduction in track lengths, but very slightly greater
> than the actual proportions involved, then this probably indicates
> pushing Gracenote just past its limits.
>
> Finally, the fact that the Simon disc also offers Hatto does suggest
> that the software did manage to identify the 'Hatto' disc correctly,
> without further intervention.
Nope. See above. It wasn't there before.
Andrew Rose
> On 2007-03-07 12:25:03 +0100, Andrew Rose <and...@pristineaudio.com> said:
>
>> Well it looks like I might have to eat my words on the subject of
>> iTunes/Gracenote. However, the Hatto/Simon discs do seem to be on the
>> very limit of the Gracenote software's abilities.
>
> Except for one thing: the Simon disc _was not in the Gracenote database_
> when this whole thing began. I checked several times, and searched
> several ways for Laszlo Simon. He _only_ showed up as being on the
> Concert Artists CD. So Bis - or Gracenote - has added the info since
> then. Basically, so what you do returns the Bis recording, and to cover
> their tracks for their tech person saying something ridiculous.
Whether the Simon disc was there a month ago is in fact irrelevant to
the tests I've performed. What I want to know is how much one has to
deviate from original timings before a CD is no longer recognised.
Unless I can prove that the way the database works means it wouldn't
recognise one as possibly the other then I'm not prepared to say it's
impossible and therefore must have been deliberate.
>>
>> I made 6 test discs and ran them through the Gracenote CDDB system:
>>
>> CD1 - Timings identical to Hatto - offers Hatto or Simon
>>
>> CD2 - Timings reduced on 3 tracks within Hatto parameters - offers
>> Hatto or Simon
>
> This suggests that they've also altered the database...
No - it suggests it falls well within their fuzzy logic.
>>
>> CD3 - Timings reduced by same number of seconds as Hatto/Simon
>> differences to same tracks - fails to recognise CD
>
> So you're saying that the timings that correspond to the Hatto recording
> don't return anything? For CD1 you are saying it does?
No - you've misunderstood this. I reduced the Hatto timings by the same
amount of seconds as the differences between the Hatto and Simon discs.
So if Hatto track 5 is 13s shorter than Simon track 5, I've used a test
disc with a track 5 that's 13s shorter than the Hatto disc.
>>
>> CD4 - Timings reduced by approx. 3/4 of Hatto/Simon time differences -
>> offers Hatto
>>
>> CD5 - Timings reduced by percentage differences between Hatto/Simon -
>> offers Hatto
>>
>> CD6 - Simon CD - offers Simon/Hatto
>
> These last three suggest that Gracenote has twiddled with the database
> again. If the timings are as different as you suggest, they shouldn't
> happen. You can try taking another CD - anything that's in the Gracenote
> dabatase with a single record - then making a test CD with slightly
> different timings to see what happens...
It looks like it'll accommodate the odd timing change of perhaps up to
just under 5%. But I'm sure it'll be a little more sophisticated than
that - the three exact matches will weigh the possibilities in favour of
a correct match, four of the timings are within 1% of their original
figures, and only three exceed a 2.7% difference.
>>
>> Finally, the fact that the Simon disc also offers Hatto does suggest
>> that the software did manage to identify the 'Hatto' disc correctly,
>> without further intervention.
>
> Nope. See above. It wasn't there before.
This is irrelevant to the tests I've done today - and as I can't wind
the clock back and check the database as it was 3 or 4 weeks ago I can't
do more than see how much one has to push the Gracenote system before it
breaks.
As far as these tests are concerned, what I think they've shown is that
the timing differences found between the Simon and Hatto discs fall just
within the current parameters of the logic used by Gracenote to identify
CDs. It was this which has surprised me, and my expectation was that the
system would not cope with these timing differences.
As such I do not think it's possible for me to continue to suggest in
interviews or online that this was _probably_ the work of someone
deliberately setting a clue.
Given that I've no intention of spending hours wasting more CDs trying
to prove a hypothesis I personally no longer think holds water, I
concede that it is entirely possible that Gracenote can identify the
Simon as a possible match when given the Hatto CD, and therefore that
this aspect of the so-called conspiracy theory can at the very least be
thrown into doubt.
Unless and until someone comes forward with evidence that they did input
this data deliberately I'll not be continuing with this particular line
of speculation or enquiry.
Andrew Rose
> On Wed, 07 Mar 2007 07:28:28 +0100, Andrew Rose wrote:
>
>> Many thanks for this - I now have a dummy CD file which matches these
>> timings precisely. As soon as I get software installed which will
>> interrogate the Gracenote database (hopefully not iTunes if I can keep
>> it away from my system!) I can start investigating...
>
>
I used WinAmp.
Andrew Rose
> Andrew Rose <and...@pristineaudio.com> - Wed, 07 Mar 2007 07:28:28
> +0100 in rec.music.classical.recordings:
>
>> As soon as I get software installed which will
>> interrogate the Gracenote database (hopefully not iTunes if I can keep
>> it away from my system!) I can start investigating
>
> The download package is much larger, but that contains a number of
> system files, most of which you are likely to have already.
> http://pucb.com/cdinfo/cd-info.htm
>
> I also recommend querying freeDB. I use EAC for that.
>
> EM
The Hatto disc is not found in the freeDB database.
Andrew Rose
Give it the Simon disc and it'll offer both. Give it the Hatto disc and
it'll offer both. Doctor the Hatto disc so it deviates as much as the
Hatto did from the Simon and it'll still offer Hatto.
The only way to prove that this was indeed not the Gracenote software
but a deliberate act is to have a full record of the Gracenote database
as it stood on 13th February 2007 with regard to both Hatto and Simon.
Without that, all is speculation, not fact.
The facts we do have:
1. Gracenote software can offer speculative matches within the timing
deviation percentages shown between the Simon and Hatto discs.
2. You couldn't find the Simon disc listed as a Bis recording a few days
later, after the story broke, and as it is now listed, suggesting that
the current entry has been adjusted or corrected.
None of which adds up to a conspiracy theory that would satisfy me.
Andrej Kluge
Unfortunately it's the only means to get the music on your iPod... :-( Or
isn't it?
Ciao
A.
Andrew Rose
Which is one device I feel I ought to know about professionally, but
have absolutely no real need or use for...
Andrej Kluge
> On Wed, 07 Mar 2007 14:19:28 +0100, Andrej Kluge wrote:
>
>>>> (hopefully not iTunes if I can keep it away from my system!)
>>>
>>> Funny... that's exactly how I feel about it. So it's not just me.
>>
>> Unfortunately it's the only means to get the music on your iPod...
>> :-( Or isn't it?
>
I got it as a xmas present -- now shall I dump it? :-)
> The best way to get your music in a mobile player (e.g. in your mobile
> phone) still is to buy a CD, rip the music and convert it to high
> quality MP3, and transfer it to your player.
That's exactly what I do (I do not buy anything -- usually DRM-infested --
from iTunes or any other MP3 online retailer). But to get these (CD-ripped)
MP3s on the iPod, you have to use iTunes, alas.
Ciao
A.
Matthew B. Tepper
to be typed in news:557sf4F...@mid.individual.net:
There is a third-party program called Rockbox which allows you to use their
player on the iPod; you then use the iPod as an external hard drive and move
your selected music tracks to it manually.
I believe Winamp may still offer the option of synching to an iPod.
--
Matthew B. Tepper: WWW, science fiction, classical music, ducks!
My personal home page -- http://home.earthlink.net/~oy/index.html
My main music page --- http://home.earthlink.net/~oy/berlioz.html
To write to me, do for my address what Androcles did for the lion
Harrington/Coy is a gay wrestler who won't come out of the closet
Matthew B. Tepper
letters to be typed in news:557tspF...@mid.individual.net:
> On Wed, 07 Mar 2007 14:19:28 +0100, Andrej Kluge wrote:
>
>>>> (hopefully not iTunes if I can keep it away from my system!)
>>>
>>> Funny... that's exactly how I feel about it. So it's not just me.
>>
>> Unfortunately it's the only means to get the music on your iPod... :-(
>> Or isn't it?
>
> That is a good reason NOT to buy an iPod. I just not like a system that
> creates dependencies between hardware, software and content. The best way
> to get your music in a mobile player (e.g. in your mobile phone) still is
> to buy a CD, rip the music and convert it to high quality MP3, and
> transfer it to your player. No strings attatched. Even LP's I bought over
> 40 years ago will work that way, as will the CD's I buy now do over 40
> years. Music lovers should not lock themselves in in digital rights
> schema's. Listeners still have some rights too in a digital world.
I've ripped many LPs and created MP3 files which play fine on my iPod.
As for DRM, I have only ever bought *one* track that way, a Pete Seeger song
that I could not find anywhere else. Of course one of the first things I did
was to strip it of DRM so that I could play it where I like.
Andrew Rose
I had a discussion a few months ago about it with a certain ex-editor of
Gramophone who warned me that it had pretty much taken over his
computer. It was enough to warn me off it, just as I was thinking I
really ought to get it, if only for research purposes...
David Fox
> Kimba W. Lion wrote:
> > Andrew Rose <and...@pristineaudio.com> wrote:
>
> >> (hopefully not iTunes if I can keep it away from my system!)
>
> > Funny... that's exactly how I feel about it. So it's not just me.
>
> I had a discussion a few months ago about it with a certain ex-editor of
> Gramophone who warned me that it had pretty much taken over his
> computer. It was enough to warn me off it, just as I was thinking I
> really ought to get it, if only for research purposes...
>
I'm very sensitive to invasive programs. iTunes is really not that
bad. It's about a 3 on a 10 point scale. It's only truly invasive
feature is that it wants to be your default audio player for MP3's and
CD's. You can tell it not to be, and I believe you can check a box
to tell it that once and for all. I have it safely installed on
several computers that I don't use for iPod synching and it quietly
minds its business like a good dormant program. It only runs when I
want it to.
DF
Andrew Rose
Thanks - that's reassuring!
Kirk McElhearn
> That is a good reason NOT to buy an iPod. I just not like a system that
> creates dependencies between hardware, software and content. The best way
> to get your music in a mobile player (e.g. in your mobile phone) still is
> to buy a CD, rip the music and convert it to high quality MP3, and
> transfer it to your player. No strings attatched. Even LP's I bought over
> 40 years ago will work that way, as will the CD's I buy now do over 40
> years. Music lovers should not lock themselves in in digital rights
> schema's. Listeners still have some rights too in a digital world.
What does that have to do with buying an iPod? There is no DRM on music
you rip yourself; you know that, right? It's not like the Zune, which
prevents you from sending music to people (or rather adds DRM to files
you "squirt" to your friends.)
Kirk McElhearn
> There is a third-party program called Rockbox which allows you to use
> their player on the iPod; you then use the iPod as an external hard
> drive and move your selected music tracks to it manually.
It's not a program, but an operating system that replaces the iPod's
OS. It's not entirely stable...
Kirk McElhearn
>>> Just one loose end that I'd like to try and tidy up - the iTunes
>>> question. Gracenotes reckon that the Hatto track durations shouldn't
>>> prevent it from identifying the CD as by Laszlo Simon, even though
>>> they're quite different in some cases.
>>
>> I've said it before, and I'll say it again. This Gracenote thing is a
>> red herring. It did not identify the Simon Bis CD, but rather the
>> Concert Artists CD having as performer Simon. CQFD.
>
> Give it the Simon disc and it'll offer both. Give it the Hatto disc and
> it'll offer both. Doctor the Hatto disc so it deviates as much as the
> Hatto did from the Simon and it'll still offer Hatto.
>
> The only way to prove that this was indeed not the Gracenote software
> but a deliberate act is to have a full record of the Gracenote database
> as it stood on 13th February 2007 with regard to both Hatto and Simon.
Well, I searched the DB several times after Hattogate broke. There were
no other Simon recordings.
>
> Without that, all is speculation, not fact.
You may say so, and it's impossible to prove otherwise.
>
> The facts we do have:
>
> 1. Gracenote software can offer speculative matches within the timing
> deviation percentages shown between the Simon and Hatto discs.
Which is interesting. I'd like to see you try some other discs and see
how minor differences turn out... You used iTunes for this, correct?
>
> 2. You couldn't find the Simon disc listed as a Bis recording a few
> days later, after the story broke, and as it is now listed, suggesting
> that the current entry has been adjusted or corrected.
No, because the Con Artists listing is still in the DB.
>
> None of which adds up to a conspiracy theory that would satisfy me.
Seems pretty clear to me. They said it "found" the correct performer,
but that is impossible, since the real disc wasn't even there...
Kirk McElhearn
> Hatto on Concert Artists, and Simon on BIS.
> The Hatto record can't be retrieved.
Hatto wasn't listed in the past; this record is still there:
http://www.gracenote.com/music/album.html/genclassical/865115b0574fa1ba9388bfa349a9038e.html
Kirk McElhearn
>>
>>> Many thanks for this - I now have a dummy CD file which matches these
>>> timings precisely. As soon as I get software installed which will
>>> interrogate the Gracenote database (hopefully not iTunes if I can keep
>>> it away from my system!) I can start investigating...
>> I use Winamp, as I don't like iTunes either.
>>
>
> I used WinAmp.
Wait, you're saying you did your tests using WinAmp and not iTunes?
Well, they are totally useless. WinAmp may indeed use acoustic
fingerprinting; iTunes doesn't. That would explain how it found
something even though the timings were different.
The only serious way for you to test this is with iTunes, since that is
what Gracenote claims "discovered" the "real" performer.
Kirk McElhearn
>>
>> Except for one thing: the Simon disc _was not in the Gracenote database_
>> when this whole thing began. I checked several times, and searched
>> several ways for Laszlo Simon. He _only_ showed up as being on the
>> Concert Artists CD.
> You write "when this whole thing began" Have you checked _BEFORE_ the
> Gramophone publication? After the publication the Gracenote DB might have
> been altered.
>
> Things we know now for sure:
>
> - Gracenote does identify the Hatto CD as Simon's
No, it identifies it as being Simon on Concert Artists.
>
> - Someone claimed his Hatto CD was identified as Simon's on Feb 13
No, they claim it showed Simon as the performer. Read the real text
that Classics Today published; it is clear.
>
> - Someone else verified that before publication (and writes the BIS
> label was on his display)
No, only Gramophone mentions Bis.
>
> - Users supply thousands of additions and corrections to the Gracenote
> DB each day.
And Gracenote does not allow them all to be added.
>
> So we better not jump to conclusions not supported by facts.
The facts are pretty clear to me...
Kirk McElhearn
>>>
>>> I made 6 test discs and ran them through the Gracenote CDDB system:
>>>
>>> CD1 - Timings identical to Hatto - offers Hatto or Simon
>>>
>>> CD2 - Timings reduced on 3 tracks within Hatto parameters - offers
>>> Hatto or Simon
>>
>> This suggests that they've also altered the database...
>
> No - it suggests it falls well within their fuzzy logic.
See my other post - if you didn't use iTunes, then your tests are worthless.
>
>>>
>>> CD3 - Timings reduced by same number of seconds as Hatto/Simon
>>> differences to same tracks - fails to recognise CD
>>
>> So you're saying that the timings that correspond to the Hatto
>> recording don't return anything? For CD1 you are saying it does?
>
> No - you've misunderstood this. I reduced the Hatto timings by the same
> amount of seconds as the differences between the Hatto and Simon discs.
> So if Hatto track 5 is 13s shorter than Simon track 5, I've used a test
> disc with a track 5 that's 13s shorter than the Hatto disc.
But with the same music?
>
>>>
>>> CD4 - Timings reduced by approx. 3/4 of Hatto/Simon time differences -
>>> offers Hatto
>>>
>>> CD5 - Timings reduced by percentage differences between Hatto/Simon -
>>> offers Hatto
>>>
>>> CD6 - Simon CD - offers Simon/Hatto
>>
>> These last three suggest that Gracenote has twiddled with the database
>> again. If the timings are as different as you suggest, they shouldn't
>> happen. You can try taking another CD - anything that's in the
>> Gracenote dabatase with a single record - then making a test CD with
>> slightly different timings to see what happens...
>
> It looks like it'll accommodate the odd timing change of perhaps up to
> just under 5%. But I'm sure it'll be a little more sophisticated than
> that - the three exact matches will weigh the possibilities in favour
> of a correct match, four of the timings are within 1% of their original
> figures, and only three exceed a 2.7% difference.
Same question - not iTunes? WinAmp? Does it use acoustic fingerprinting?
>
>
>>>
>>> Finally, the fact that the Simon disc also offers Hatto does suggest
>>> that the software did manage to identify the 'Hatto' disc correctly,
>>> without further intervention.
>>
>> Nope. See above. It wasn't there before.
>
>
> This is irrelevant to the tests I've done today - and as I can't wind
> the clock back and check the database as it was 3 or 4 weeks ago I
> can't do more than see how much one has to push the Gracenote system
> before it breaks.
No, but I checked very carefully. The only change re Hatto and Simon is
the addition of the Bis recording, which wasn't there in the beginning.
There are still the same Hatto recordings, and still the same Simon on
Con Artists recording listed.
>
> As far as these tests are concerned, what I think they've shown is that
> the timing differences found between the Simon and Hatto discs fall
> just within the current parameters of the logic used by Gracenote to
> identify CDs. It was this which has surprised me, and my expectation
> was that the system would not cope with these timing differences.
>
> As such I do not think it's possible for me to continue to suggest in
> interviews or online that this was _probably_ the work of someone
> deliberately setting a clue.
>
> Given that I've no intention of spending hours wasting more CDs trying
> to prove a hypothesis I personally no longer think holds water, I
> concede that it is entirely possible that Gracenote can identify the
> Simon as a possible match when given the Hatto CD, and therefore that
> this aspect of the so-called conspiracy theory can at the very least be
> thrown into doubt.
>
> Unless and until someone comes forward with evidence that they did
> input this data deliberately I'll not be continuing with this
> particular line of speculation or enquiry.
Same question - what softare?
Kirk McElhearn
>>> Hatto on Concert Artists, and Simon on BIS. The Hatto record can't be
>>> retrieved.
>>
>> Hatto wasn't listed in the past; this record is still there:
>>
>> http://www.gracenote.com/music/album.html/genclassical/865115b0574fa1ba9388bfa349a9038e.html
>>
> Laszlo Simon / Franz Liszt -12 Etudes d'execution transcendante
> Label: BIS Records
> Year: 1987
The above link shows, for me:
Laszlo Simon / Franz Liszt -12 Etudes d'execution transcendante
Label: Concert Artists
Year: 2007
I'm sayng that this recording is still there, as is, now, the Bis
recording (which wasn't there when Hattogate broke).
Kirk McElhearn
>>>>> Hatto on Concert Artists, and Simon on BIS. The Hatto record can't be
>>>>> retrieved.
>>>>
>>>> Hatto wasn't listed in the past; this record is still there:
>>>>
>>>> http://www.gracenote.com/music/album.html/genclassical/865115b0574fa1ba9388bfa349a9038e.html
Laszlo
>>>>
>>> Simon / Franz Liszt -12 Etudes d'execution transcendante
>>> Label: BIS Records
>>> Year: 1987
>>
>> The above link shows, for me:
>>
>> Laszlo Simon / Franz Liszt -12 Etudes d'execution transcendante Label:
>> Concert Artists
>> Year: 2007
> I can't help it if your browser settings are wrong, but that record is
> not on Gracenote.
My browser "settings" are wrong? How do you explain that you and I,
then, are getting different info from the DB? Do you want me to send
you a screen shot? Send me one of what you see... For the URL above.
Because there's something even more fishy if different people are
seeing different things on the same URL.
BTW, do note the exact text from the Classics Today article:
"When he attempted to transfer this recording via his computer using
iTunes, the track listing revealed the artist as Laszlo Simon, who
recorded the same music for BIS."
It does not say that it showed the Bis disc, but rather Simon as the
performer...
Kirk McElhearn
> Your browser caches this retreival from a database. That is wrong. Try to
> hold the shift key while pressing reload, that might help.
My browser cache is always off. I made a PDF of the page.
Interestingly, now it's only showing the Bis recording. It looks like
they just updated the DB (or rather, they updated it recently, and it
took until now for it to get to me - I'm sure they use distributed
servers, which would explain why you were seeing a different page than
I.)
>
>> BTW, do note the exact text from the Classics Today article:
>
> Read the Gramaphone article, who checked the Gracenote behaviour (and
> wrote the label BIS was displayed on the display) and checked the files
> are identical by sending them to Andrew.
Actually, Gramophone didn't check anything; the original info did not
come from them, but from Classics Today, who passed the info on to
Gramphone, who then went public before Classics Today.
>
> We have discussed this earlier, so we better stop.
Yes, and we discussed it before, when there was no Bis recording for
Simon listed... You might have checked then, but perhaps you forgot?
Kirk McElhearn
> y use distributed servers, which would explain why you were seeing a
> different page than I.)
>>
>>> BTW, do note the exact text from the Classics Today article:
>>
>> Read the Gramaphone article, who checked the Gracenote behaviour (and
>> wrote the label BIS was displayed on the display) and c
One more thing - Andrew said he was seeing _two_ records. So now one
has been deleted...
Matthew B. Tepper
letters to be typed in news:45eeeb33$0$27392$ba4a...@news.orange.fr:
> On 2007-03-07 16:33:46 +0100, "Matthew B. Tepper" <oyž@earthlink.net>
said:
>
>> There is a third-party program called Rockbox which allows you to use
>> their player on the iPod; you then use the iPod as an external hard
>> drive and move your selected music tracks to it manually.
>
> It's not a program, but an operating system that replaces the iPod's
> OS. It's not entirely stable...
My error. And I note from the Website that they still don't have a version
of it which works on the 80GB 5.5Gen iPod with video.
Steve de Mena
Technically, the Zune doesn't add anything or
modify the file being "squirted", it just adds the
DRM restriction to the database entry for that
file. They would have run afowl of the Creative
Commons licensing people if they were physically
modifying their files.
Someone will probably come up with a way to
transfer that file back to a PC in its original
un-DRM state....IF the Zune remains in the
marketplace long enough. (I think Microsoft
relased Zune way before the software was finished
and everything was worked out).
Steve
Steve de Mena
> On 2007-03-07 13:18:06 +0100, Chel van Gennip <ch...@vangennip.nl> said:
>
>>>
>>> Except for one thing: the Simon disc _was not in the Gracenote database_
>>> when this whole thing began. I checked several times, and searched
>>> several ways for Laszlo Simon. He _only_ showed up as being on the
>>> Concert Artists CD.
>> You write "when this whole thing began" Have you checked _BEFORE_ the
>> Gramophone publication? After the publication the Gracenote DB might have
>> been altered.
>>
>> Things we know now for sure:
>>
>> - Gracenote does identify the Hatto CD as Simon's
>
> No, it identifies it as being Simon on Concert Artists.
I see Bis.
http://www.gracenote.com/music/album.html/genclassical/865115b0574fa1ba9388bfa349a9038e.html
Laszlo Simon / Franz Liszt -12 Etudes d'execution
transcendante
Label: BIS Records
Track Title
1. Preludio
2. Molto vivace
3. Paysage
4. Mazeppa
5. Feux follets
6. Vision
7. Eroica
8. Wilde Jagd
9. Ricordanza
10. Allegro agitato molto
11. Harmonies du soir
12. Chasse-neige
Andrew Rose
>>
>> Read the Gramaphone article, who checked the Gracenote behaviour (and
>> wrote the label BIS was displayed on the display) and checked the files
>> are identical by sending them to Andrew.
>
> Actually, Gramophone didn't check anything; the original info did not
> come from them, but from Classics Today, who passed the info on to
> Gramphone, who then went public before Classics Today.
Wrong. Jed Distler is not Classics Today. Neither is he Gramophone. He
writes for both.
He e-mailed both editors at the same time with the same e-mail.
Gramophone acted on it, both by commissioning an independent listening
test and by commissioning me to check it out technically. Then they
published.
I can't speak for CT and how they handled the story, but the bottom line
is that both had exactly the same information at exactly the same time
from the same source. That Gramophone turned it around, authenticated it
and published it is to their credit, IMHO.
Andrew Rose
> On 2007-03-07 13:36:42 +0100, Andrew Rose <and...@pristineaudio.com> said:
>
>>>>
>>>> I made 6 test discs and ran them through the Gracenote CDDB system:
>>>>
>>>> CD1 - Timings identical to Hatto - offers Hatto or Simon
>>>>
>>>> CD2 - Timings reduced on 3 tracks within Hatto parameters - offers
>>>> Hatto or Simon
>>>
>>> This suggests that they've also altered the database...
>>
>> No - it suggests it falls well within their fuzzy logic.
>
> See my other post - if you didn't use iTunes, then your tests are
> worthless.
I'm talking about the Gracenote database, as used by both. Does the
software do anything other than pass on track timing information to
retrieve possible matches? If so, what?
Andrew Rose
No - as you well know. What difference does this make? You fail to explain.
Look, Kirk, I really am not going to get into a public argument about
this. I've used software which interrogates the Gracenote database with
track timing information in order to try and substantiate the theories
I've made public, and for which I was taken to task in the Wall Street
Journal, and which have been printed and broadcast elsewhere, that
suggest it's likely that the database was doctored.
I said this without doing any research to substantiate my claims. It
seemed to make common sense - in hindsight that is not enough.
I've now done my own research which suggests to me that I could have
been wrong. I chose to make this public as I was carrying it out, but it
was entirely independent and for my own interest, having spent an
afternoon stating the same theory for French TV.
What is your big problem with this? If you wish to take over and print
the same theory, go ahead. I did some tests which throw doubts into my
mind. That's all. It proves nothing except that I can't prove my own
original theory, when I thought I would be able to do so.
I'd rather talk about facts that I can verify than guesswork, and that's
all the iTunes/Gracenote theory is.
Kirk McElhearn
>> See my other post - if you didn't use iTunes, then your tests are worthless.
>
> I'm talking about the Gracenote database, as used by both. Does the
> software do anything other than pass on track timing information to
> retrieve possible matches? If so, what?
Yes, as I said in my other post, if one is using acoustic
fingerprinting, it's a totally different animal.
Kirk McElhearn
> Technically, the Zune doesn't add anything or modify the file being
> "squirted", it just adds the DRM restriction to the database entry for
> that file. They would have run afowl of the Creative Commons licensing
> people if they were physically modifying their files.
From what I understand, they have. CC or someone has raised this issue.
(I don't know what the results are because the Zune seems to have
fizzled...)
Kirk McElhearn
"Steve Scherf, Co-Founder and VP of Service Development and
Architecture for Gracenote, submitted a major correction to this post.
Apple iTunes does not use Gracenote's acoustic recognition, and
instead, uses "plain-old CD recognition" to identify CDs. "
http://blog.wired.com/music/2007/02/song_recognitio.html#more
The quote is in the update to the article; the journalist originally
thought that iTunes was using acoustic fingerprinting. I'm still trying
to find out if WinAmp does...
Kirk McElhearn
> Regarding Andrew's tests, IMHO they are not conclusive. They cannot
> be, because Gracenote CDDB changes all the time, and it has been
> altered with respect to this particular item.
Yes, and the software does matter too...
>
> Keep in mind that CDDB can be searched textually:
>
> <http://www.gracenote.com/music/index_old.html>
>
> When the story broke I did so. There was the following entry:
>
> -----
> Laszlo Simon / Franz Liszt -12 Etudes d'execution transcendante
> Label: Concert Artists
> Year: 2007
> Artist Info : Joyce Hatto
As did I, and I posted about it several times. Anyone could have
checked at the time to see that.
> -----
>
> but no entry for Simon without Hatto or mentioning the BIS label. At
> present, this same entry has been changed to:
>
> -----
> Laszlo Simon / Franz Liszt -12 Etudes d'execution transcendante
> Label: BIS Records
> Year: 1987
> Artist Info : László Simon, Piano
Yes, which is interesting, because this change was made yesterday (at
least I saw it yesterday after other posts here.)
> -----
>
> (See <http://www.gracenote.com/music/album.html/genclassical/
> 865115b0574fa1ba9388bfa349a9038e.html>. Keep in mind that information
> is cached at various points along the way, so it is not impossible
> that sometimes the old data might be retrieved.)
My thought is that they have distributed servers, since they get
queries from around the world. Such servers, like DNS servers, can take
up to 48 hours to get their content updated. So the change could have
been made any time in the past 72 hours or so...
>
> I have no way of knowing whether this entry was of the "Hatto" or the
> Simon CD, but it seems reasonable to hypothesize that the original
> Simon CD was not in CDDB; the only CDDB entry was for the "Hatto" CD,
> and someone, intentionally or not, inserted Simon as the artist for
> that entry. (Had the initial entry been of the original Simon CD,
> there would have been no reason to enter the label as CA instead of
> BIS, or to add Hatto's name.) But I have no proof other than my
> recollection. Only Gracenote may be in a position to confirm or deny
> it, and I strongly doubt they will do either.
No, because they came out smelling like their system was "smart", and
they want to prove that it is. Of course, removing the original record
is pure cover-up...
Note that when Andrew was testing yesterday, he claimed to be seeing
_two_ records at some time, suggesting that the change was being made
right around then.
>
> IMHO, the only thing we can do at this point is to determine whether
> the differences between Simon/BIS and Hatto/CA fall within Gracenote's
> fuzziness or not. To do so, we need, (a) the precise Hatto/CA listings
> from the actual CD -- posted by Walter, (b) the precise Simon/BIS
> listings from the actual CD, (c) the precise listings for a recording
> of the Etudes transcendantes listed in Gracenote CDDB (Berezovsky,
> Jando, Oldenburg, etc.). A simple tool which does this is discid
>
> <http://discid.sourceforge.net/>
>
> but there are plenty of others.
>
> We need (c) because fuzziness is very likely dependent on the number
> and lengths of tracks, so it's best to determine it on a CD as close
> in layout as possible. Having (c), one can build a dummy CD and
> variations on it, and thus determine CDDB's fuzziness. Then one may
> compute whether the differences between Hatto/CA and Simon/BIS fall or
> not within it. It's best to run the fuzziness tests with iTunes
> (rather than WinAmp), to eliminate other variables. On a Mac, the
> tests do not require burning actual CDs -- one can use a tool such as
> Roxio Jam to build mountable audio CD images, which, as far as iTunes
> is concerned, are indistinguishable from actual CDs.
Yes, it is essential to use iTunes, which was claimed to have been used
to "unmask" this hoax. Other software may use acoustic fingerprinting.
As I posted yesterday, Gracenote stated that iTunes doesn't use
acousting fingerprinting, but nowhere on the Gracenote site does it say
which software _does_ use it. Frankly, I don't think they want to say,
because there are some privacy issues around this technology (people
being afraid that it can spot music that has been obtained illegally).
Kirk McElhearn
> And don't forget, even if Winamp does use the audio fingerprinting, you
> still need to know whether the data for that particular CD included
> audio fingerprinting, since that technology hasn't been applied to
> everything in the Gracenotes database.
My guess is that, if acoustic fingerprinting is used, there's a chance
this CD would have had it. If you notice the record, the data is
supplied in Gracenote's "new" classical music format (artist name /
work name). They have done this for a number of CDs, mostly new ones or
those recently added to the database, and it is possible that these CDs
are fingeprinted as well. It may also depend on which software is being
used to upload the info; the existence of a label name for the original
record of this disc shows that it was not iTunes, since iTunes doesn't
let you enter label names.
Andrew Rose
>> In article <45ef3573$0$27385$ba4a...@news.orange.fr>, kir...@gmail.com says...
> still need to know whether the data for that particular CD included
> audio fingerprinting, since that technology hasn't been applied to
> everything in the Gracenotes database.
Winamp simply checks the CD durations. I haven't investigated the audio
fingerprinted concept in this particular context, but bear in mind that
to do so after inserting a CD the following must take place:
1 - The entire CD needs to be ripped to the hard drive - taking several
minutes as well as disc space.
2 - This file needs to be then loaded and analysed. The data processing
required is going to take another minute or two at least if they're
looking at the acoustic profile of the audio.
3 - This profile, which will have to be more complex, and therefore a
larger file, will then have to be submitted for checking - a much slower
upload and cross-checking process than simply looking at CD durations.
Thus any software in which you simply pop a CD into the player and it
quickly checks against CDDB is almost certainly doing none of the above,
and is going purely on CD track durations.
It's the latter that I was testing for - with a series of CDs with null
content (i.e. total silence) where the manipulation of track timings was
the only alteration made. Forget any hokum about acoustic fingerprinting
- if the software appears almost instantaneous all it can be doing is
checking durations.
Note again - I've proved nothing other than the fact that the Gracenote
time recognition system, as it was configured yesterday, _is_ capable of
recognising a CD which deviates from the original to the same degree as
the Hatto did from the Simon.
Kirk McElhearn
> It's the latter that I was testing for - with a series of CDs with null
> content (i.e. total silence) where the manipulation of track timings
> was the only alteration made. Forget any hokum about acoustic
> fingerprinting - if the software appears almost instantaneous all it
> can be doing is checking durations.
That's a valid assumption.
>
>
> Note again - I've proved nothing other than the fact that the Gracenote
> time recognition system, as it was configured yesterday, _is_ capable
> of recognising a CD which deviates from the original to the same degree
> as the Hatto did from the Simon.
What gets me is the fact that there are so many CDs, and so many
possibilities that you'll have similar timings, that this doesn't make
sense.
Add to that the fact that _there was no record for the Bis CD when this
broke_, and your tests only proved that as Gracenote changed the record
(some time yesterday or the day before) you managed to hit the Bis
record.
It's still neither iTunes nor Gracenote that "discovered" the Bis CD
when someone checked the "Hatto" CD - for the simple reason that
Gracenote didn't have a record for that CD. I'm not the only one who
checked when this all broke.
Andrew Rose
> But, in Gramophone's haste, they mislaid some facts, e.g., they didn't
> mention that the original tip came from Ventura, and instead made it
> appear that Distler made the discovery all on his own, IIRC. Also, I
> don't remember that their story mentioned iTunes, specifically.
>
AFAIK they didn't know that the original tip came from Ventura - it
wasn't a name which I'd heard or seen until I read it in the CT
coverage. Distler did have the Hatto disc, checked it himself, and
reported his findings.
The original Gramophone text, posted on RMCR on 15th Feb, ran as follows:
"Several days ago, another Gramophone critic decided to listen to a
Hatto Liszt CD, of the 12 Transcendental Studies. He put the disc into
his computer to listen, and something awfully strange happened. His
iTunes player identified the disc as, yes, the Liszts, but not a Hatto
recording. Instead, his display suggested that the disc was one on BIS
Records, by the pianist Lazlo Simon. Mystified, our critic checked his
Hatto disc against the actual Simon recording, and to his amazement they
sounded exactly the same."
I don't have a copy of his e-mail to Gramophone/CT so do not know
precisely what detail was enclosed about who did what, when. I do know
that Distler did his own research having been "tipped off by a reader" -
and that at the time he wanted to keep his name out of the story, hence
the slight muddying of the information in the first line.
> But of course, since they hired you, it's understandable that you'd
> want to praise their coverage.
I'm not sure how clarifying the situation counts as praising anyone's
coverage. If Distler didn't want to be directly associated with the
Gramophone news release then that was his choice, and the story had to
be adjusted to accommodate that.
I was merely correcting the notion that the discovery had anything to do
directly with Classics Today. Both publications received the same info
from Distler - who contributes to both of them and had been contacted
directly by "a reader" - at the same time. What they chose to do with it
was up to them.
aleksios
said:
> We know the Hatto CD can be identified as the Simon CD by
> gracenote
Actually, we don't.
> We know someone discovered his Hatto CD was identified as a Simon CD
> on feb 13
Again, we don't know. It is entirely possible that the CD was
recognised correctly, without any fuzziness, and that the Gracenote
entry contained "Simon" as artist, instead of, or in addition to,
Hatto. What actually happened is even more cloudy, because the text
quoted by Andrew mentioned BIS, and, (1) I do not recall "BIS" in the
Gracenote entry at that time; (2) iTunes does not use/display labels.
> It is not so likely Gracenote spends a lot of time checking
> contributions for the millions of CD's in their DB.
Indeed. But -- at this point -- we are talking about only one, or at
most two, records, which have been for at least two weeks something of
a minor cause célèbre. It is entirely within the realm of practical
possibility that Gracenote may have fiddled with it (or them) since
the story broke.
If someone posts the data I requested in a previous message (the
actual CD data, not data from CD-R copies, MP3s, or on-line services),
I can run the tests and post both computations and raw results.
However, there are two issues to keep in mind:
(1) Tests can only tell us what could have happened, not what _did_
happen.
(2) The entire issue may be moot. My recollection -- which seems to
coincide with that of Kirk, and perhaps others -- is that there was
only one entry for this CD in the Gracenote CDDB on 2007/02/13. That
entry contained "Simon", "Hatto", and CA, but not "BIS"; hence, it may
be considered reasonable to assume the entry was of the Hatto CD. So
there was no involvement of Gracenote's fuzziness at all; and someone
must have added both "Hatto" and "Simon" to that entry when submitting
it to Gracenote.
--Alex
aleksios
> [...] Hatto always quicker where
> the times have been changed: [...]
I compared the values posted by Walter for "Hatto" to the values for
Simon from BIS and the values for Jeno Jando from Naxos (as control).
You can see the comparison here:
<http://aleksios.netfirms.com/misc/table.html>
Note that these values show "Hatto" tracks to be usually longer, not
shorter, contrary to Andrew's findings. I assume this is due to
omission or inclusion of pre-track pauses. (Alternatively, BIS may not
be listing the correct values.) But, as explained here
<http://ftp.freedb.org/pub/freedb/misc/freedb_howto1.07.zip>
discID is calculated from frame offsets and total CD duration, so a
comparison based on track duration can only be an approximation. FWIW,
compared to my previous test, the value for average absolute deviation
tends to confirm Andrew's conclusion that the Hatto/Simon difference
is somewhere at the limit of Gracenote CDDB's fuzziness. However, this
is very rough. Only raw data from the physical CDs can provide support
(including the actual discIDs) for a conclusion one way or the other.
--Alex
Andrew Rose
The tracks durations you have for the BIS recording are incorrect - in
some cases by a large margin! (Not your fault...)
The precise durations are as follows, as taken from the eMusic download
- in my experience these appear to be straight rips from the CDs:
1. 0:49.057
2. 2:17.325
3. 5:06.155
4. 7:53.182
5. 3:46.194
6. 5:23.186
7. 4:49.802
8. 5:23.526
9. 9:48.173
10. 4:27.180
11. 9:39.186
12. 5:44.084
Quite how the BIS website listing manages to be so far out I don't know...
Andrew Rose
> On Thu, 08 Mar 2007 17:00:33 +0100, aleksios wrote:
>
>> My recollection -- which seems to coincide with that of Kirk, and
>> perhaps others -- is that there was only one entry for this CD in the
>> Gracenote CDDB on 2007/02/13.
>
> the Gramophone publication, you are the first suspects of any
> manipulation of these records.
>
It's worth pointing out something very curious indeed that I discovered
yesterday (incidentally during a long phone discussion on the subject
with Kirk).
A website search of Gracenote's database returns two Hatto entries - the
Liszt and the Chopin Etudes.
However, if you query it using a software client with either of the two
Hatto Godowsky Chopin Studies CDs, it 'correctly' identifies them. If
you then search the website for the Godowsky the Hatto does not show. In
other words there are entries in the database which only appear as disc
matches, and not as searchable results.
All this suggests to me that the whole CDDB/iTunes question is likely to
prove exceptionally difficult, if not impossible, to answer conclusively.
Kirk McElhearn
> It seems likely that the Hatto CD was discovered with the CDDB database,
> and for this discovery it was sufficient that the BIS Simon CD was in the
> database and it is not uncommon for CD's to be listed in the DB.
>
> There are no facts at all supporting speculations about a discovery
> because someone planted specific information in the CDDB. No planted
> information was needed. If there had been planted information, it seems
> logical that information would have named BOTH pianists, and not only
> Simon. So until there are real facts, there is no room for speculations.
Yet again... When the story broke, several people, including myself,
checked the Gracenote database, and the Simon Bis recording was not
listed. Now, it is true that after what Andrew and I discovered
yesterday - that the web search and software search return different
results - there is a possiblity that it was indeed there. However, I'll
stick with what Classics Today reports, that the _performer_ was Laszlo
Simon, not that iTunes returned the disc as being Laszlo Simon's Bis
disc. Because that fits with the record that was visible on the
Gracenote web site until two days ago...
Kirk McElhearn
> Quite how the BIS website listing manages to be so far out I don't know...
Different labels treat pauses differently; some add them into the track
lengths, and others don't.
I've often seen track lengths (shown in software) that differ from
what's on the CD info... The software lengths are correct, because they
are calculated from the catalog info.
aleksios
said:
> The precise durations are as follows, as taken
> from the eMusic download - in my experience
> these appear to be straight rips from the CDs:
Unfortunately, we cannot assume these are the actual figures. Here's
why:
(1) The initial pause has to be at least 150 frames (2s), but it's not
at all uncommon for it to take values between 180 and 182 frames. No
rip can provide this datum.
(2) Ripping software treat track pauses differently. Some (e.g., EAC)
give the user control over whether the pause will be included or not,
others (e.g., iTunes) do not. Moreover, some ripping software also
shorten digital silence at the beginning and end of tracks, even when
it's part of the track rather than the pause.
(3) Ripping and encoding themselves can also slightly alter the number
of frames.
Consequently, the only data that can lead to a reliable test is raw
frame counts read directly from the original physical audio CD, as I
mentioned earlier. Unfortunately, it looks like no-one reading this
thread actually owns a copy of the Simon CD.
However, I did put in the track lengths posted by Andrew (truncating
fractions of seconds) here:
<http://aleksios.netfirms.com/misc/table2.html>
With the new figures, the deviation is 1.7, which I believe to be
beyond Gracenote's fuzziness; it would suggest that Gracenote could
not have mistaken one CD for the other. However, once again, absent
the actual CD frame offsets, the figures are not reliable. Moreover,
the parameter I'm using (track and CD lengths normalised average
deviation relative to the Simon BIS CD) may not be indicative of
Gracenote's true fuzziness.
--Alex
aleksios
said:
> [...] if you query it using a software client
> with either of the two Hatto Godowsky Chopin
> Studies CDs, it 'correctly' identifies them. If
> you then search the website for the Godowsky the
> Hatto does not show. In other words there are
> entries in the database which only appear as
> disc matches, and not as searchable results.
> [...]
That's a possibility; but the flaw in your tests is that you used
Hatto. Gracenote is hardly unaware of Hattogate, so any test which
assumes black box behaviour with respect to a Hatto entry must be
flawed. IOW, what you are seeing is not necessarily Gracenote CDDB's
usual behaviour. It's not impossible that Gracenote may have blocked
browser access to some Hatto entries, while still allowing access to
them to CDDB2 protocol clients.
The fact that little Johnny doesn't pick his nose when he knows he's
being watched doesn't mean he's not picking his nose when he thinks no-
one is watching. So CDDB's behaviour must be tested with CDs of other
artists, which are not linked to the Hatto affair.
--Alex