Is it possible to taint the AR DB?

Topic: Is it possible to taint the AR DB? (Read 2491 times) previous topic - next topic

0 Members and 1 Guest are viewing this topic.

Is it possible to taint the AR DB?

2010-05-18 23:55:59

- I've been holding off on submitting my AR results using EAC because i noticed something. When ripping some of my very old CDRS, i saw that popular albums that should definately be in the database were coming up as 'track not in AR database' instead of coming up as 'in-accurate' like most CDRS. Does that mean if I submit my results i could possibly taint the AR database because the DB will think its just a different album or a different pressing or something? Is there someone behind the scenes who screens for this, or is the DB itself able to screen through this kind of stuff based on artist and album info? I'm probably being overly paranoid, but i think caution is always best if your actions can affect other peoples results. And even if there are measures in place like that, if i did have a bad rip and the disc isnt actually in the database then couldnt that bad rip be submitted as the first submission for that album?
- I recently came across a situation that may be linked to this kind of problem. When i first ripped my 'Massive Attack - Mezzanine' album it had some errors and came up as in-accurate against a 200 confidence level. So i borrowed a burn of it from my friend to attempt to replace it. When i ripped my friend's album, it came up as accurate but against only 1/1 confidence. When i listen to the apparently 'accurate' album at 1 confidence i seem to be hearing subtle quiet clicks in the background. Now this may be just the way the album is due the artist possibly using samples or something, but i don't remember hearing that before. I tried to tell myself that the odds of getting the exact same in-accurate rip as someone else were crazy, but the situation just seems odd. Unless somehow the rip i got at confidence 1/1 is a different pressing from the 200, and my pressing is from a bad gold disc or something like that.

Any input would be great.

Is it possible to taint the AR DB?

Reply #1 – 2010-05-19 00:48:14

Discs are identified by AR using the disc's TOC. Any deviation from the original disc's TOC such as the first track starting at a different place (due to the presence of a 00 index), the presence of a data track or tracks having different lengths will appear to AR as being an entirely different disc. Differences in track lengths can occur quite easily depending on what software you've used and how it was configured. It's common for burning programs such as Nero to add two seconds between tracks by default. It's not uncommon for ripping programs to remove additional silence or rip all tracks short by one frame as well.

The database has no way of knowing that submissions are from CD-R, let alone whether they may contain errors. Because it works based on positive matches, however, the odds that two people will encounter the exact same error because of a scratch or glitch in the ripping process are practically non-existent. If someone is checking an original factory pressed disc against the database, the presence of a CD-R rip with the same disc ID or a submission with errors* makes no difference.

(*) Ignoring errors that may be consistent due to a drive with buggy firmware, a bug in the ripping software, a pressing with a manufacturing defect or any combination of the three.

Is it possible to taint the AR DB?

Reply #2 – 2010-05-19 01:14:49

So I can't taint the database then?

What about the situation with the massive attack CD? Any ideas as to what could have happened there?

Is it possible to taint the AR DB?

Reply #3 – 2010-05-19 12:16:47

Quote from: Skybrowser on 2010-05-19 01:14:49

So I can't taint the database then?

What about the situation with the massive attack CD? Any ideas as to what could have happened there?

Your rip may have come up accurate (1) because your friend ripped the same disc. Anyone else who rips a copy of that title will not match, so no worries.

As to the clicks, are you referring to the vinyl surface noise from the sampled records Massive Attack uses. Do they loop in concert with musical samples?

Funny thing is, at a conference presentation on the history of mp3 given by Jonathon Sterne, he demonstrated different bitrates/encoding (I'd have to check my notes but he played .wav, 128 mp3, and maybe 256 mp3 and 256 aac) and used that album as the test track. I was able to correctly identify all four samples blind, through a rented PA with plastic speakers, based in part on how the surface noise from the samples was rendered (also drum dynamics). Could have been dumb luck, and certainly not statistically significant.

Is it possible to taint the AR DB?

Reply #4 – 2010-05-24 01:50:15

I can ask my friend. By i'm almost 100% certain he didnt. And if he did rip the disk, it wouldnt be using a program that has AR.

Is it possible to taint the AR DB?

Reply #5 – 2010-05-24 02:11:56

I see nothing to worry about. When I ripped my Danzig I original CD (not CDR), I was first surprised to see that such a good album (even if not specially popular) wasn't in the database ... I discovered later that there is in fact another much more popular pressing than mine. My rare pressing is still now only AR2 (which is perfect for my needs anyway) ... so just wait & see. It does happen that some pressings are particulary rare, even on CD that should be popular enought to be in the database. My guess is that my Danzig I CD is a rare local french pressing (not even a special edition with any bonus) with a slight burning difference & as neither Danzig nor accuraterip are specially popular in france, it was painfull just to get to AR2. Specially as 1 of the 2 results is mine ...

Edit:
Well it seems that Danzig I is more popular than I thougth
From Wikipedia:

Quote

Danzig is the band's best-selling full length album and was certified Gold in the U.S. in 1994.

The very popular pressing in the database that doesn't match my CD must be the US pressing, simply.

Edit2:
I just re-checked this rip, in my case the rarety of my pressing is likely due to the unusual pregap length ... I forget to check the Pregap of the popular pressing when I had it at hands.

Code: [Select]

[Verification date: 05/05/2010 03:24:44]
[AccurateRip ID: 00102505-008061de-7f09990a] found.
Pregap length 00:00:27.
[CTDB TOCID: Xo6an7lz2uJq_8rFE8ZFgn_KAJE-] disk not present in database.
Track	[ CRC    ] Status
 01	[dabb8238] (02/02) Accurately ripped
 02	[4064d3df] (02/02) Accurately ripped
 03	[0df039b3] (02/02) Accurately ripped
 04	[8b6321d3] (02/02) Accurately ripped
 05	[826fcf47] (02/02) Accurately ripped
 06	[3bb5f848] (02/02) Accurately ripped
 07	[3406fb76] (02/02) Accurately ripped
 08	[52ff99ac] (02/02) Accurately ripped
 09	[40eb2b33] (02/02) Accurately ripped
 10	[304dfcfa] (02/02) Accurately ripped

Track Peak [ CRC32  ] [W/O NULL] 
 --  100,0 [DC99B9B9] [FD9E5B65]           
 01   84,6 [8BB9F6B6] [823BE7C7]           
 02   82,2 [3F06C4A1] [EC3185FB]           
 03   77,0 [F178E187] [B775F272]           
 04   84,7 [6B8D0CF4] [211425FA]           
 05  100,0 [6A99B155] [15A330C6]           
 06   73,4 [E79B9E44] [F4500E7F]           
 07   93,2 [1F956EBC] [78B91B3A]           
 08   73,5 [6B8453C5] [B4D906AC]           
 09   76,6 [60C4C871] [20574DD4]           
 10   71,9 [7FD9A2A2] [55ED90D3]

Notice