I am finding out that there are Audio CDs from different pressings, with the same TOC and with an offset built in

So a new method of Offset detection is required. It is quite a simple concept - as well as having a database of accuraterips from discs, also have a database of Offsets for drive makes.

Now any disc from AccurateRip can be used to find the offset, if it matches what is in the drive database then all is well - configure accuraterip. If it is a new drive or or the offset does not match what is in the database then do not accept that offset until a 2nd recognised (or Key) disc gives the same offset.

I am currently writing this lot into AccurateRip Test #3

Flawless?

Wait... I must check my own offsets.
There might be a problem : I've got three CDs showing a sudden cut at the end of last track, all three matched, and the offset of the cut was 12 samples away from EAC's reference, if the common write offset in SatCP's for Yamaha 4416 and 8416 burners stands for 6416 too.
Andre Wiethoff also reported that when he was detecting the first offset, the one he chose, same on 6 CD, was "the one that came out the more often", and that I actually came across three CDs pressed in a factory using a different offset that the one from which his CDs came.
So other offsets may easily match.

Now that I've got a more reliable reference (the Teac E540, 6 matching reports to SatCP), I must see if these three CDs are indeed offsetted.

my yamaha 6416 scsi burner does not produce the same offset, it's different every time, not a very reliable drive in that respect

Done.

I have no CD from EAC or your list.
So I did as Andre ( http://pageperso.aol.fr/lyonpio2001/offset.htm ), I looked at the end of the wavefiles, to see if the end of the recording was visible.

I found four CDs, three were matching. The read offset correction they give is 14 samples higher compared to the all the ones reported at SatCP for
-The Yamaha 4416 to 8416 series (5 matching reports), granted the 6416 have the same write offset.
-The Sony DDU1621 1.6 (one report, certainly from Halcyon)
-The Teac 540E 3.0 (6 matching reports)
-The Yamaha CRW3200 (1 report, 1.0b instead of 1.0d).

The fourth one deviates by +2183

However, that doesn't imply that we can find the same error on two different CDs, that would be an error compared to another version of both.

To clear this, one should gather all possible wrong offset detections, and see if some wrong numbers come more than once.

for example, my liteon 24102b burner turns in an offset of +12 everytime, but check out the log for the yamaha using the same disc

Finding Offset for F:\ [YAMAHA - CRW6416S ] - 15 Tracks [FreedB ee12630f, Added 002f6368, Multi 020ef596]


Track 1 - CRC matching 92946f89 [confidence 3]
Offset match at 168 samples (672 bytes)

Track 2 - CRC matching 4935c98f [confidence 3]
Offset match at 166 samples (664 bytes)

Track 3 - CRC matching 6af9e8a3 [confidence 3]
Offset match at 169 samples (676 bytes)

Track 4 - CRC matching 7eee2721 [confidence 3]
Offset match at 167 samples (668 bytes)

Track 5 - CRC matching 378b05c5 [confidence 3]
Offset match at 169 samples (676 bytes)

Track 6 - CRC matching e0a7fcc0 [confidence 3]
Offset match at 165 samples (660 bytes)

Track 7 - CRC matching 00000000 [confidence 0]

Track 8 - CRC matching 610f64df [confidence 2]
Offset match at 169 samples (676 bytes)

Track 9 - CRC matching 25092c8e [confidence 3]
Offset match at 169 samples (676 bytes)

Track 10 - CRC matching c2e5a2ab [confidence 3]
Offset match at 169 samples (676 bytes)

Track 11 - CRC matching 65cb4630 [confidence 3]
Offset match at 169 samples (676 bytes)

Track 12 - CRC matching 581b7835 [confidence 3]
Offset match at 169 samples (676 bytes)

Track 13 - CRC matching 66327f5f [confidence 3]
Offset match at 169 samples (676 bytes)

Track 14 - CRC matching 15252e52 [confidence 3]
Offset match at 169 samples (676 bytes)

Track 15 - CRC matching 20885767 [confidence 3]
Offset match at 165 samples (660 bytes)


Not all offset values calculated matched, this most likely means your disc used to configure AccurateRip has scratches, please try a different matching disc.

yes, this is the very reason that I note the bad discs that I have ripped when I email spoon latest rips. If I get something like that above, I know that the disc has scratches, and so I can tell him not to include that disc in the database.


Question: Are you keeping the crc database information and the FreeDB information completely independent of each other? IE, the FreeDB track and artist naming info that I am choosing may be incorrect, but does that matter for crc verification? If it does, imo, it shouldn't.

@mmortal03

FreedB is not used for disc Identification.

As you are ripping twice, first on one drive then on another the database will self clean - any of your discs with scratches will be automatically removed because they don't match.

The CRC matching show that your CD is in good state.

The Yamaha 6416S has no accurate stream, it means that its offset is variable. I used to have one, and got exactly the same results as you : the offset varied in a window 6 samples large.
The write offset (+13) seemed much more stable than the read offset, but if I remember well, I saw it vary once (among about 10 CD burned).

would it not be worth it to keep the info anyways so that at least good tracks would be able to be matched by others? track by track rather than disc by disc

It is done on a track by track basis, and because the CRC of track detection is just 1 CD sector that small amount of data is not normally affected by scratches (if it is then that disc will not have all matching Offset values so it will not be used).