The ONLY way to be guaranteed correct ripping?
Reply #17 – 2003-05-02 21:50:57
There is no known way to be sure that you have 100% accurately got the data from the CD correctly, especially if the CD is scratched. I've read up on this a bit now, and I disagree with the above claim. *) If you have a drive that reports C2 errors with a 100% reliability (oh please God, make it exist!) *) And you get no C2 errors when reading the CD *) GOTO ::ReadGoodCD ... [nitpicking mode] I disagree with the above claim. If you find errors in the following estimation feel free to correct me For each chunk of audio data containing 24 bytes there's 4 bytes of C1 and 4 bytes of C2. IIRC C1 as well as C2 information is capable of locating+correcting 1 error (not only of detecting if the chunk is unchanged or not), so there's redundancy in C1/C2 of at least 50% (maybe I'm wrong with this as I'm not very familiar with the mathematics of error detection/correction algorithms). So there are (at most) 2^(2*4*8*50%) = 2^32 possible C1/C2 combinations that have to cover 2^(24* =2^192 possible data chunks. So each C1/C2 combination gives correct checking results for 2^(192-32)=2^160 data chunks. If you change a 24 byte chunk randomly, the probability to get correct C1/C2 checking results is ~1/1^32 Not taking into account all the mistakes made so far, to get a probability of 0.5 = 50% of getting an undetectable C1/C2 error you need on average N randomly changed 24 byte data chunks with (1-1/2^32)^N=0.5, so N=2977044471 which is 112.5 hours of audio if every 24 byte chunk is randomly changed. This is not 100% security, even with a drive that reports all detectable C2 errors (and corrects all correctable C1 errors)!!! But it's enough security for me to be satisfied with this faulty calculation. For you maybe ripping two copies of the same CD and comparing the results might give enough security. [/nitpicking mode] About the reading twice vs. use C2 information issue: Usually, if an error that the drive can't correct using C1 info occurs there'll be - maaaany more errors nearby and - a random behaviour in extraction results of those errors. So in 99.9xxx % of dirty/scratched/... CDs erroreous positions will be noticed by reading twice and comparing the results. This has been tested and is result of the experience of many EAC users. If you believe testing yourself is better than trusting you can do it like this: Take a brand-new CD with perfect surface and ripp it - C2 on test & copy or use whatever method you trust most ATM (or a CD that you have burned yourself - make sure that you have the original .wav files stored at a safe place), Than scratch it, paint some lines on it with a thin marker, eat a hot dog and wipe your fingers on the data side afterwards, ... whatever you want. Then rip with EAC secure mode A. C2 enabled, Test & Copy B. C2 disabled, "Drive caches ..." checked, Test & Copy, check if T & C CRCs match and compare the results to the original with some sound editor's wave substraction or EAC's wave compare. If the result for a track is "no errors occured" and CRCs match there should be no differences compared to the original. If there are differences and/or CRCs don't match you know that you can't trust the settings you tested.