IPB

Welcome Guest ( Log In | Register )

AccurateRip confidence - what does it really say?
OliP
post Jan 7 2013, 02:00
Post #1





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



I have a CD RIP with EAC V1.0 beta3
Checked with CUETools 2.1.4

CUETools gives me here:
.
.
03 [f41e2a14|bc31a59f] (39+08/88) Accurately ripped

So thats means, this track matches with 39 in AR V1 and with 8 in AR V2.
Makes 47 matches alltogether. But there are 88 entries!
Are these 41 other rips, rips with errors? Or maybe different pressings?


Well !
My logical thinking tells me that read errors always occure on different places on CDs right?
So erroneous rips do always have different checksums and none of them matches another, right?
So if I have a confidence 1, its most likely that I have a bit-identical copy. Otherwise me and the other user, we both must have a read error at exactly the same place, so that our checksums are identical, and this is very unlikely right ??
So when I have AR confidence 1 and the database also has many other results I do not match with, its most likely that my rip is correct, and all the others are wrong, (or different pressings) right ???


Iīve never really understood this...
But:
If I have a confidence of only 1, my rip is at 99,9999999999999999999999 % a Bit-identical copy, right?


This post has been edited by OliP: Jan 7 2013, 02:09
Go to the top of the page
+Quote Post
 
Start new topic
Replies (1 - 15)
spoon
post Jan 7 2013, 09:22
Post #2


dBpowerAMP developer


Group: Developer (Donating)
Posts: 2758
Joined: 24-March 02
Member No.: 1615



The answer is different pressings, also you only need a confidence of 1 (as long as you did not submit this result)


--------------------
Spoon http://www.dbpoweramp.com
Go to the top of the page
+Quote Post
OliP
post Jan 7 2013, 15:12
Post #3





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (spoon @ Jan 7 2013, 09:22) *
also you only need a confidence of 1 (as long as you did not submit this result)

could also mean, that we both have an erroneus rip!
So is ist it correct, what I assume, that read errors nearly always appear on different places, and thats why erroneus rip checksums as good as always differ each other??
And therefore a match with just one other person means a perfect rip. Notice well, not 100% sure, but with a probability of 99,99999%.

I nowhere found an explanation about that in german language, so now thats just a logical attempt to understand it !
For a match with one person does not automaticaly mean an errorfree rip!
Can someone technical versed, give me a confirmation for my thinking?

This post has been edited by OliP: Jan 7 2013, 15:19
Go to the top of the page
+Quote Post
spoon
post Jan 7 2013, 16:07
Post #4


dBpowerAMP developer


Group: Developer (Donating)
Posts: 2758
Joined: 24-March 02
Member No.: 1615



Anything is possible, even if improbable.

The way AccurateRip works once a disc has been verified by a 2nd person (confidence of 2), then it takes up that 'slot' in the database, to add to this slot (such as a different pressing) requires 2 verifications. That means ripping errors tend to be excluded from the database (as generally speaking ripping errors are unique to your disc + also to some extent your drive).

So this leaves a false negative chance, and the odds against this are quite huge (you have more chance of being struck by lightening), over the years AccurateRip has processed 143M discs, I am sure that possibly one of the tracks on one of those rips had a false positive, but unless you are ripping 143M discs I would not worry about it.

This post has been edited by spoon: Jan 7 2013, 16:16


--------------------
Spoon http://www.dbpoweramp.com
Go to the top of the page
+Quote Post
OliP
post Jan 8 2013, 16:23
Post #5





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (spoon @ Jan 7 2013, 16:07) *
So this leaves a false negative chance, and the odds against this are quite huge

I donīt understand this.


Itīs correct that erroneous results (nearly) always differ each other.
So when I have an erroneous rip, I will not match with anyone.
Right?

Thatīs all I wanted to know

This post has been edited by OliP: Jan 8 2013, 16:33
Go to the top of the page
+Quote Post
Engelsstaub
post Jan 8 2013, 16:49
Post #6





Group: Members
Posts: 574
Joined: 16-February 10
Member No.: 78200



Es ist sehr unwahrscheinlich. (It is very unlikely.)

If your rip contains errors it is extremely improbable that another submission will contain those exact same errors.

An exception is this: if you illicitly download someone else's rip and use a program like XLD or CUETools to verify this rip, it can return an AR result of at least 1. This is obviously because that same rip was likely already submitted to the AR database. (We know the moral of that story.)


--------------------
The Loudness War is over. Now it's a hopeless occupation.
Go to the top of the page
+Quote Post
pdq
post Jan 8 2013, 17:03
Post #7





Group: Members
Posts: 3450
Joined: 1-September 05
From: SE Pennsylvania
Member No.: 24233



Another possibility, though I don't know how likely this is: If there is a flaw in the glass master for a CD such that multiple copies have exactly the same incorrect data, then the rips will be verified by AR.
Go to the top of the page
+Quote Post
greynol
post Jan 8 2013, 18:23
Post #8





Group: Super Moderator
Posts: 10345
Joined: 1-April 04
From: San Francisco
Member No.: 13167



In addition, defective drives and issues with the ripping software can result and, in all reasonable certainly, have resulted in the population of the AR database with active* hashes for errant rips.

(*) active meaning confidence two or greater; confidence of one if there are no other validated submissions for any particular disc ID.

This post has been edited by greynol: Jan 9 2013, 02:00


--------------------
Breath is found in plots and DR figures.
Go to the top of the page
+Quote Post
OliP
post Jan 8 2013, 21:13
Post #9





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (pdq @ Jan 8 2013, 17:03) *
Another possibility, though I don't know how likely this is: If there is a flaw in the glass master for a CD such that multiple copies have exactly the same incorrect data, then the rips will be verified by AR.

So thats a production mistake at a certain CD charge, meant so?
But will happen only on rare occassions, right?

Ok, now Iīm going to understand the kidding-like sentence of spoon - "anything is possible"

QUOTE (greynol @ Jan 8 2013, 18:23) *
In addition, defective drives and issues with the ripping software can result and in all reasonable certainly have resulted in the population of the AR database with hashes for errant rips.

yes, but that falls in the category erroneous rips, where itīs so much unlikely to have exactly the same error on another machine, and get a confidence for that.
OK
Go to the top of the page
+Quote Post
greynol
post Jan 8 2013, 21:20
Post #10





Group: Super Moderator
Posts: 10345
Joined: 1-April 04
From: San Francisco
Member No.: 13167



You're misunderstanding. By defective I meant a flaw in the design that may affect an entire line of drives sharing common firmware. I hope the software portion of my comment doesn't need similar qualification.


--------------------
Breath is found in plots and DR figures.
Go to the top of the page
+Quote Post
OliP
post Jan 8 2013, 21:22
Post #11





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (pdq @ Jan 8 2013, 17:03) *
Another possibility, though I don't know how likely this is: If there is a flaw in the glass master for a CD such that multiple copies have exactly the same incorrect data, then the rips will be verified by AR.


And - in that case, I would get an AR verification, but have ripped an error, that was now implemented in my wav-file.
So it could be possible hear that clearly in the wav file, but not when playing the CD on a CD-Player that corrects such errors to unsensable!
But my EAC also has an error-correction...
Go to the top of the page
+Quote Post
greynol
post Jan 8 2013, 21:26
Post #12





Group: Super Moderator
Posts: 10345
Joined: 1-April 04
From: San Francisco
Member No.: 13167



EAC's "error correction" is only limited to synchronization issues. Otheriwise asking the drive to continually re-read data and pick what comes up most consistently hardly constitutes error correction.

Regarding concealment, (most?) computer drives will handle it the same way as the old stand-alone players. Is there even a difference these days between computer drives and the transports in standalone players?

This post has been edited by greynol: Jan 8 2013, 21:37


--------------------
Breath is found in plots and DR figures.
Go to the top of the page
+Quote Post
OliP
post Jan 8 2013, 21:51
Post #13





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (greynol @ Jan 8 2013, 21:20) *
You're misunderstanding. By defective I meant a flaw in the design that may affect an entire line of drives sharing common firmware. I hope the software portion of my comment doesn't need similar qualification.


A defective drive charge.
But thatīs also very seldom.

What about the ripping software?
A certain CD ripped by 10 persons with an unprofessional or wrong configured software will give me confidence 10 when I use the verry same software or wrong configuration and the same CD?


OK we donīt have to count beans here.
A confidence of 1 means an errorfree rip for 99,999999999 %, if Iīm sure, that itīs not my own past submission.
Go to the top of the page
+Quote Post
greynol
post Jan 8 2013, 22:11
Post #14





Group: Super Moderator
Posts: 10345
Joined: 1-April 04
From: San Francisco
Member No.: 13167



This is an old conversation which I'm really not interested in having yet again. I've analyzed tens of thousands of rips for myself and on the behalf of others and have seen my share of anomalies.

So have fun blindly guessing at probabilities. smile.gif


--------------------
Breath is found in plots and DR figures.
Go to the top of the page
+Quote Post
OliP
post Jan 8 2013, 22:52
Post #15





Group: Members
Posts: 30
Joined: 6-January 13
Member No.: 105636



QUOTE (greynol @ Jan 8 2013, 22:11) *
This is an old conversation which I'm really not interested in having yet again. I've analyzed tens of thousands of rips for myself and on the behalf of others and have seen my share of anomalies.
So have fun blindly guessing at probabilities. smile.gif

analyzed rips? share of anomalies?

Oh, a german speaker does not understand this sentence.

Well, I also want to close this conversation.
Confidence 1 means a bit identical copy, for 99,99999999 % probability.
Now I just want to have the information of the other thread, if the AR DB can distinguish results.
So then, a confidence of 1 for a seldom/rare CD is OK, when Iīm sure that I was not the person who caused this in the past.

This post has been edited by OliP: Jan 8 2013, 22:53
Go to the top of the page
+Quote Post
greynol
post Jan 8 2013, 23:48
Post #16





Group: Super Moderator
Posts: 10345
Joined: 1-April 04
From: San Francisco
Member No.: 13167



Something along the lines of this as one of many examples that can be found online:
http://www.digital-inn.de/exact-audio-copy...-related-3.html

Take from it what you will. I'm not going to spend any more time on the matter. smile.gif

This post has been edited by greynol: Jan 9 2013, 01:53


--------------------
Breath is found in plots and DR figures.
Go to the top of the page
+Quote Post

Reply to this topicStart new topic
1 User(s) are reading this topic (1 Guests and 0 Anonymous Users)
0 Members:

 



RSS Lo-Fi Version Time is now: 28th December 2014 - 14:46