For those who are new to AccurateRip - it is my effort to create an eventual online database, much like freedb but this one contains CRCs of peoples CD Rips so that others can assertain if their rip is accurate. It is my understanding that this is the only way to achieve 100% AccurateRips. Should this prove successful the technology will be opened up for everyone and will probabbly appear in CDex and EAC.

Developments:

The system has been made more resistant to being infected by corrupt results by making it a requirement that a 'keying' ie recognised disc for finding offsets has to have a confidence of 2, that is the result were submitted by two independant people.

A Logfile is created when ever an offset is found.

Options menu >> Accurate Rip menu (will also tell if a disc is in dB before ripping, and the last accuraterip result).

On Options is a clear offset button to reset the offset of a drive to unconfigured.

====================
Please visit here for full details:

http://forum.dbpoweramp.com/showthread.php...d=6471#post6471

Anyone who has over 100 discs in their collection who would like to take part (and do not have a matching known disc) please email dbpoweramp@dbpoweramp.com I will make sure you get a disc. Thanks.

It sounds an interresting project to me.
Is it possible for a "Various Artists" CD to have each track have their own Artist/Title parameters ? Or the Artist will be "Various Artists" in the database ?

It has nothing to do with Disc names, they only represent what is in the database - as it uses freedb in an automatic manor some of the names are wrong as it picks the first name it comes across, but as I said it is only for representing what is in the database the names are not used in the database (a hash of the TOC is used).

Sorry but what you are saying is definitely not clear to me !
Is it a FreeDB replacement ?

No.

Just so I make sure I'm getting this right...

Any disc that has been read by EAC, for example, will always give the same CRC value? So if user X has a cd, and user Y also has the same cd read on two completely different systems. They will both get this number? Provided that the cds are from the same pressings, countries and other things that may give a different read.

Am I close, in a basic way?

Guys, the idea is to build a database which lets you test whether your rip of a CD is accurate (and compute offsets?). It computes some things from your ripped audio and sees if those match the numbers in the database of known good rips of audio. It's a little like db.etree.org's shns in circulation database, but for officially released albums (each of which has many variants).

Spot on buzzy and MachineHead.

It also computes the offsets of drives as this is required to get accurate rips.

Yes, but how??

Considering the fact that most rips will be done without offset-corrected drives,, or with badly offset-corrected drives (the most easy and therefore most frequently used way to correct offsets is consulting a database... the database says +50000? OK, let's use +50000 then. And in the end hundreds of people are using this offset value, and possibly corrupt your database this way)

I hope you found a solution but I doubt there is one

A good point and one which is the real danger to the database. For a minute forget how people find offsets by looking at a published list, that is the old way and you are right there is a large percentage who think they have set up a correct offset but really it is wrong.

AccurateRip imposes this about offsets: It has to find the offset - no manual inputting, it can find an offset once a disc in the database is inserted. Until then, that computer does not partake with AccurateRip and to make sure that that offset it right, the confidence of the disc in the database has to be two or more, that is those results came from two seperate people.

If that was not enough the system in 'Test #3' will constantly check the offset and warn if it is not right.

How will it react when someone wants to rip a different pressing ? Will it correct the offset between the reference and the CD inserted ? Will it rip it as it is, warning about the offset, but ensuring a proper rip, or will it tell that the CD is not in the database ?

A different pressing as far as I can tell they have different track lengths so appear as different discs. Take this disc 'Eminem - The Marshall Mathers LP' US release:

http://www.freedb.org/freedb_search_fmt.ph...ock&id=0f10ef12

and UK:

http://www.freedb.org/freedb_search_fmt.ph...ock&id=1e121c13

or 'Fatboy Slim - You've Come a Long Way, Baby' US:

http://www.freedb.org/freedb_search_fmt.ph...ock&id=990e8a0b

and UK:

http://www.freedb.org/freedb_search_fmt.ph...ock&id=9f0e8c0b

BTW the fat boy slim UK release has a Fat bloke on the front, the US release is just a picture of LPs on a shelf...

I was talking of pressings with different offsets. I'll be able to check the two Lisa Gerrard CDs I've got from France and England when my computer is fixed (I get data corruption now), but I checked track 1 when I got it, and they were bitwise identical, just offsetted.
According to user reports not finding constant offsets with EAC CDs, it seems that 80% at least of the released pressings of a given CD are offsetted from each other.

Are we talking sample offsets or whole CD sectors?

got the cd today...any use in re-ripping with different drives? Got a couple drives in this machine and a couple in the other machine. Shall I re-rip with the other ones too? Or would it not give you any data you're interested in?

Yes, it will help verify your own results, as long as the drives are a different make. Rip using one drive - send the Results to me, then re-rip with other drive and send the results. (it is important the results are sent seperately).

yeah, all four are different brands/types of drives. Went a lot quicker than I thought, so I'm sure I'll be able to find some time to do it with the other drives as well. At this point I've only been doing it with the one drive, so I don't have any mixed results in the file you'd like. 1:26 am here so I'm off to bed, I'll finish up the handful of discs I have left tomorrow sometime and send you that file and I'll work on it with the other drives as I get time.

<edit> I assume you'd still like that file to be deleted before I go ahead with a new drive, correct? And if so is there some way to access the updated database so that I can verify the rerips? Or will I just be leaving the file intact in this case?

Send me the results - delete results, I will send you an updated database with your results embedded.

Spoon-

I tried this out. But it did not offer to configure my drive (from instructions on your site). And it seems there was no .bin file anywhere. Did search the harddrives but came up with no results.

Is there something I might have done wrong in configuring that would cause this to not be set up correctly?

Anyway, this sounds very interesting and I'd like to see if it works for me.

Mike

It sounds like none of your discs matched what was in the database (different pressings), having it offer to find the offset is the first step.

OK. According to the list I have two for sure, and maybe a third. Have to check on one. I only tried Oasis so far. I'll see if the other known one (from the list) gives a different result.

Radiohead worked this time.

Sent the results.

EDIT: This disc does have this bonus track. It just is not listed on the cover. Sorry for confusion. Read it with EAC and Media Center, and it shows up with in both of these. Results should be what you are looking for.

But how does it find the offset? How does it work, I mean?


Be warned... for EAC it happened with various people that the first + second CD give e.g. +9 and then another 2 CDs give +10.

I assume that if you have a cd that it reccognizes, but it actually isn't the one you need, and if it finds a bad offset this way, you will know when you compare your crcs to the ones in the database...either that, or IT already IS checking the crcs a couple times of the disc when it verifies the offset...this is different than EAC EAC doesn't have a database of crcs to check to as well for verifying one's offset test cds.

The offset detection in AccurateRip is supposed to be spot on - finding the offset not just over one track, but every track on the disc.

I tried EACs auto offset detection on a bunch of my discs and it was a little jumpy even though it recognised the discs. I am not sure offset detection in EAC was done though - I should really get in touch with Andre and ask for a few matching discs out of his database so tests can be run on those discs.

it got the same result, +12, as EAC with my liteon 24102b burner anyways.

yeah, same here, but I have also had it to reccognize discs that were actually not the correct pressing and were offsetted, thereby finding the wrong offset. I think what spoon is saying is that this won't happen with his program.

Could someone who knows Andre approch him and ask how EAC finds the offests? how it identifies a known disc and the mechanism used to find the offset on that known disc.

I think AccurateRip should be immune to different pressings, at least I hope so - there are two things in its favour - It recognises a disc on the TOC, different pressings should have different TOCs, even if they do not to find the offset it will search up and down about 2 sectors for 1 sector it knows the CRC for (50 seconds into the track), if the disc was different it shouldn't match.

I have dropped Andre an email...

What if someone finds out, that his rip is not accurate (regarding to the info in the database)? What should he do? Shoot himself? What if the rip in the database is not accurate? Who wins?

I don't see why different pressings would necessarily have different TOCs. The same album issued worldwide will be a loss less copy of the masters sent to the factories, but each factory will release a CD with a different offset.
At the beginning, mismatching offsets were blamed on the fact that EAC users were using american CDs while EAC's database was based on european pressings. But we realized that the offsets mistmatched even in europe, even in Germany. It is even common (see EAC mailing list) to find CDs that have the same pressing number on the inner ring (same factory etc), but different offsets.

Thinking about it, I realized that it was still possible to find the right offset using several CDs because there was more possible offsets (several thousands) than possible pressings (several ones). Thus when two offsets from two different CDs are matching, it is much more probable that they are both the same pressing as the one referenced in EAC's database, than they have both the same offset.
I should retry to detect manually my offsets (http://pageperso.aol.fr/Lyonpio2001/offset.htm). When I had the Yamaha 6416S, I found the same offset three times, that looked different from the ones reported at SatCPs for all the other drives of the same kind (4416, 8416...).
If they are still different from SatCP's on my new drives, it means that either SatCP's database is wrong (due to users relying on wrong results to calculate their offsets), either there are matching offsets that are different from the reference one. In the later case, we'd have to trade reference CDRs in order to build the database.

I am ready to trade reference cd's when you are...I frankly don't think its such a bad idea. Someone could possibly even make a small amount of donation money or even charge a small fee (which would be fine) and burn hundreds of these to cheap CMC Media and send them out to whomever needed one in the community. CMC media would do fine, as it is the cheapest usually, and these test cds just have to last long enough to test one's drive . I guess he or she could charge more for Taiyo Yuden Test cds!

Depending how EAC is working out the offsets - if it is pretty much the same method I will try to locate two such CDs - one of them would have to be Andres CD, another one off a user.

I've nearly got two of them.
Lisa Gerrard - The mirror Pool (CAD 5009 CD - England)
Lisa Gerrard - The mirror Pool (7243 8 40592 2 4 - France)

Copy image and create cuesheet, then compare wav gives :
/1476 repeated samples at 0:00:01.062
Different samples 1:08:13.552 - 1:08:13.853 / 1:08:13.586 - 1:08:13.886
0:00:00:040 longer /

Thus they are offsetted by 1476 samples, the binary content is the same, exept the very end.

FC cuesheets gives :



Thus the TOC only differs in tracks 12 and 14.

Nearly there

As a user, you identify a known disc by reading a long list of CDs for which the data has been contributed, finding a title that you have, and checking to see if it matches the number on your pressing. Repeat 100 times to find a couple matches. The identity of the disc is not detected by EAC automatically for drives with unknown offsets.

I believe the database was created by users with known offsets who submitted data for CDs in their collection.

Presumably if you have a known pressing for which you have data, calculating the offset is not difficult (assuming away the issue Pio noted about different pressings having the same ID number).

just noticed, the test rip method still creates directories in the output directory. Wondered why I had a crapload of directories in there and then I noticed there were no files.

The next release of AccurateRip (test #3) will solve that small bugget

I have been in contact with Andre, he was most helpful, but it leaves me in two minds if there are discs out that that can upset accuraterip.

A new site accuraterip will be on line in the next few days - the goal for this site is 3 fold:

Provide AccurateRip database / resource
All drive - automatically collected and updated drive offset database
A resource for beating CopyProtected CDs - free exchange of information between developers.

yeah, I noticed that as well and posted something about it on the dBpoweramp board last week.



I like the second one, I hadn't even thought of using AccurateRip to do that. Great idea, because with AccurateRip, it can be verified.